Managing overwhelming levels of model inventory and ensuring sustainability of processes

Managing overwhelming levels of model inventory and ensuring sustainability of processes

By Chris Smigielski, Director, Model Risk Management, EverBank.

Chris, can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is? 

Sure, I’ve been in banking and financial services for over 25 years.  I started my career in Asset Liability Management (ALM), focused on market risk modeling, financial model development and model validation.  Since then, I’ve worked in several consulting roles where I’ve been able to work directly with clients of various sizes and complexity.  Currently, I am with TIAA Bank’s Enterprise Risk Management Group. My current role is Director of Model Risk Management where I’m responsible for all aspects of our Model Risk Management (MRM) program.  Our MRM program is in a fairly mature place now and we’re seeing some of the challenges from a growing model inventory as the company grows and adapts to changes in the retail and regulatory environments.

You will be presenting at the upcoming Risk Americas 2018 to discuss managing overwhelming levels of model inventory and ensuring sustainability of processes. Why is this a key talking point in the industry right now?

Models continue to provide information to influence and drive key decisions and actions.  For example, models are used to determine capital adequacy, assess various financial risks, produce financial statement information, and drive key business decisions. Models are used across the company for a variety of applications, and the potential is that model errors can produce significant losses, cause reputational damage, and influence poor business choices. A key talking point to illustrate that is the growth of machine learning and other methods and techniques which are generating greater numbers and a greater frequency of new models and better models – all which must be validated and governed – likely causing stress to resources and program efficacy.

What are the main points for consideration for maintaining a complete model inventory and why is this so challenging across institutions?

A detailed and comprehensive model inventory a vital tool for MRM to be able to understand and manage risk.  On the surface, maintaining a complete model inventory would not seem to be a challenging objective but it inherently is.  As an illustration, models can be built by the business unit to solve an immediate problem, which can create time gaps from development-to-identification and inventory-to-governance.  There should be a proactive process for model identification with education and awareness to support that process.  This discovery or notification process helps ensure our model inventory remains complete.

How can institutions ensure sustainability of their process for model inventory, managing models and tracking the model lifecycle?

It is really important to define a comprehensive MRM program and follow it.  The known model inventory, model validation and governance work is our business-as-usual (BAU), but the introduction of several models following a business initiative or a regulatory change can be disruptive to that cadence.  To ensure sustainability or program efficacy, we have a process to risk rate models because guidance is clear that materiality matters; higher risk models are then validated more often than lower risk models. We do use a limited-scope approach in some of these cases and also try to evaluate models in groups or batches if the methodology or approach is somewhat similar.  This is an illustration of tools or approaches that give us flexibility to pace our work, maintain a high quality work product, and ensure sustainability of our MRM Program.

 How do you see the risk and regulations industry evolving over the next 6-12 months, particularly in related to the current and upcoming political landscape?

Whether you are a DFAST bank or a larger CCAR bank, Iwould say that there appears to be more focus on model development and data lineage / data quality, in particular.  Regardless of the change in regulatory expectations or even in politics, our effort is to manage risk well at our institution while meeting those regulatory mandates.


 Hear insights like this and more at the Annual Risk Americas Congress…