Case study
Portfolio and risk model migration after an acquisition
Led the move of a ~1M-account credit card portfolio and its risk models into the acquiring bank’s cloud data environment.
- Client
- PNC
- Period
- Sep 2021 – Mar 2022
- Role
- Consultant, Data Science & Strategy at Management Solutions
- Where
- Pittsburgh, on-site and remote
- ~1Maccounts migrated
Context
On June 1, 2021, PNC acquired BBVA USA. The Credit Card Director I had been working with moved to the acquiring bank and, because of our performance on the previous engagement, the work was extended to PNC. I relocated to Pittsburgh.
Challenge
Move a portfolio of about one million credit card accounts, together with the risk models that ran it, from the acquired bank’s systems into a new cloud data environment, and give the business reporting it could rely on afterwards.
My role
I led the migration, reporting to the Credit Card Director and managing one team member.
What I did
- Migrated the portfolio and its associated risk models from the acquired bank’s systems to PNC‘s cloud data environment, using SAS for data extraction and transfer.
- Built the post-migration reporting framework.
- Categorized the portfolio’s customers by segment.
Alongside the migration: business development
This period added a new dimension to my work: winning new business for the firm.
- ACV Auctions. A new client for the firm. I prepared the value proposition and the AS-IS / TO-BE assessment that won the engagement, then delivered the digital implementation of the company’s origination and account-management models: an origination model assigning a credit grade to each customer, credit risk models for account management, behavioral alert models flagging customers whose performance deviated from the average beyond a defined confidence interval, and automated decision-support processes, including inference and past-due prediction models.
- CEMEX. Proposed and presented a statistical inference model to identify the variables driving cement production quality, building on earlier manufacturing analytics work.
- IFRS 9 product. Developed an internal expected-loss engine for banks, based on PD, LGD and EAD, to estimate required loan-loss reserves.
Results
- Portfolio and risk models migrated to the acquiring bank’s cloud data environment, with a post-migration reporting framework and customer segmentation in place.
- A new client won for the firm through a proposal I prepared, followed by delivery of its credit models.
Technologies
- SAS
- Cloud data environment
- Statistical origination, behavioral and inference models