Article

Bank Director: Taking model risk management to the next level – better business decisions from data consistency across models

Sept. 14, 2021 · Authored by Ivan Cilik, Sean Statz

As appeared in Bank Director

A financial institution’s data is one of its most valuable resources. Banks constantly collect data on their loans, deposits and customer behaviors. This data should play a key role in how financial intuitions manage their risks.

Yet, developing a data strategy can be seen as too complex based on the sheer amount of data an institution may have, or as an unnecessary burden if the objective is solely to use the information to satisfy regulatory requirements. But a holistic data strategy can enhance value across all model risk management (MRM) platforms, both for regulatory and strategic purposes. On the flip side, being inconsistent or not updating data and inputs in a timely manner can lead to inaccurate or inconsistent results. Executives need to continually update and review information for consistency; if not, the information’s relevancy in assessing risk across various platforms will decrease.

Currently, the most common data strategy approach for banks is using individual tools to measure risk for regulatory purposes. For instance, financial institutions are required to calculate and monitor interest rate risk related to their balance sheet and potential movements in future interest rates. Typically, one team within the institution extracts data and transfers it to another team, which loads the data into an internal or external model to calculate the various interest rate profiles for management to analyze and make decisions. The institution repeats this process for its other models (credit, capital adequacy, liquidity, budgeting, etc.), adjusting the inputs and tools as needed. Often, banks view these models as individual silos — the teams responsible for them, and the inputs and processes, are separate from one another. However, the various models used to measure risk share many commonalities and, in many aspects, are interdependent.

Integrating model risk management processes require understanding a bank’s current data sources and aggregation processes across all of its current models. The first step for executives is to understand what data is currently used across these platforms, and how your organization can utilize it other beyond just checking the regulatory box. In order to enhance data quality, can one data extract be used for multiple platforms? For example, can the same loan-level data file be used for different models that use similar inputs such as asset liability management (ALM) and certain CECL models? While models may utilize some different or additional fields and inputs, there are many fields — such as contractual data or loan prepayment assumptions — that are consistent across models. Extracting the data once and using it for multiple platforms allows institutions to minimize the risk of inaccurate or faulty data.