Collecting and storing quality data for CECL model requirements

Collecting and storing quality data for CECL model requirements

By Shannon Kelly, SVP, Director, Model Risk Management, Zions Bancorp

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

My experience is in model development and validation across different risk types, including credit, market and operational risk. My current role is the Head of Model Risk Management at Zions Bancorp.

You will also be discussing data requirements under CECL, what are some of the challenges with collecting and storing data for CECL models?

Data for CECL requires a history over at least one economic cycle. Collecting and storing the necessary internal portfolio drivers over 1-2 economic cycles to build econometric forecasts of default and loss over the reasonable and supportable period is a challenge for many banks and limits the predictive power of the forecast models. Additionally, data elements on prepayments are challenging to compile over 1-2 economic cycles for reliable prepayment modelling.

How can institutions ensure quality of data from origination through the life of loan for CECL?

Given that historical data has challenges, gathering data on a go forward basis will involve oversight from Data Governance on the data gathering and collection processes, data testing and reconciliation controls and data quality reporting and remediation on an ongoing basis. Additionally, the most valuable current and potential future data elements to be used in the CECL estimation process need to be identified so that their quality can be ensured in this process.

CECL looks to fundamentally overhaul practices when implemented, how can this work be leveraged for more effective risk management?

CECL can be used for what If scenario analysis to provide management with information on potential losses under different plausible scenarios for risk management and pricing.  Additionally, CECL models can be used to make strategic decisions by analysing the allowance levels required for different portfolio compositions.