Credit Risk Modeling and Analytics Infrastructure

Oakleaf was retained to provide modeling and analytics services to support Stress Testing practice for a mid-size bank, which included the development of applications and tools for credit risk models, aiming for an integrated, efficient, and controlled model execution and analytics environment.

This model infrastructure included a Credit Demand Estimation Tool, implemented in SAS for aggregating credit risk forecasts and estimating capital reserves across the Client’s entire loan portfolio, an SAS plotting and tabulating engine allowing modelers to create and customize charts and tables for the Stress Testing reports, and an Interactive Reporting Tool developed with Excel/VBA, regularly used in management meetings to interactively analyze modeling results and decide strategies from model overlays to capital reserves. A documentation Automator was also utilized, developed with Word/VBA to intelligently update graphs and tables in the CCAR reports, and a Credit Loss Projection Tool (CLPT) that was developed with an Excel/VBA/SAS framework and provided capabilities of integrated execution for all credit risk models, data management, controlled run configuration, and tracking on execution status and metadata.

Our Approach

The project consisted of multiple sub-tasks over a period of several years to build a suite of applications and tools to support model execution, data analysis, and reporting. All the solutions were proposed after a comprehensive analysis of the regulatory requirements, the Client’s project status, technology environment, timeline, and budget, and were implemented in accordance with the Client’s IT governance.

The Client’s existing Credit Demand Estimation Tool (CDET) was created with MS Access, which was prone to error (consisting of 100+ SQL queries and manual steps) and low performance. Oakleaf assessed the status of the technology environment, and redesigned and developed the tool using SAS, which significantly improved the performance, integrity, and robustness. Also, with the introduction and implementation of the structured and modular coding, the response time for updated processing logic and tool maintenance was significantly reduced. These improvements provided the capabilities of integrated and controlled executions and analysis of complex scenarios.

After analyzing the Client’s regulatory requirements, Oakleaf proposed and developed the Credit Risk Model Execution Platform (with an Excel/VBA/SAS architecture) within two months to meet the deadline for production execution. The execution platform integrated the previously decentralized, modeler-dependent model execution process, and provided the application/UI layer that allows controlled configuration for regular and overlay runs, data management, and tracking of execution process and metadata. Combined with the practice of structured and modularized model implementation, the platform significantly mitigated the client’s credit model execution risk and amply met the regulatory requirements for the execution control.

Results

Oakleaf supported the Client’s analytical research and statistical analytics, and developed a SAS plotting and tabulating engine to automate the generation of all charts and tables used in the credit risk CCAR Stress Testing reports.
A superior feature was introduced with the automation of the process of embedding model results into documentation and formatting for end user presentation and consumption.
An Excel/VBA-based data visualization tool was developed that enables executives, managers, and modelers to interactively analyze modeling results and decide strategies from model overlays to capital reserves.
Oakleaf developed the Credit Risk Model Execution Platform (with an Excel/VBA/SAS architecture) within two months to meet the deadline for production execution, significantly mitigating the client’s credit model execution risk and meeting the regulatory requirements.