Company: USAA
Company background: USAA was founded in 1922 when 25 Army officers decided to insure each other's vehicles. Over time, their tiny organization has grown to serve millions in the military community. Our mission is to empower our members to achieve financial security through highly competitive products, exceptional service and trusted advice. We seek to be the #1 choice for the military community and their families. We're among the leading providers of insurance, banking, investment and retirement solutions to members of the U.S. military, veterans who have honorably served and their families.
Contact: Nick Tornabene
Title: Executive Director - Loss Reserves and Forecast Infrastructure
Country: United States of America
Award Category: Innovative Problem Solver
Tell us about the business problem you were trying to solve?
I typically do not like to write things about myself; however, I think it is important to encourage more SAS customers to think of the SAS relationship as a partnership rather than just some vendor providing software.
There are many SAS products being utilized at USAA, yet I’m going to focus on the Risk Stratum platform, Current Expected Credit Loss (CECL) and Stress Testing content packages, and the Model Implementation Platform (MIP). Specifically, I will focus on some of the feedback I’ve provided as a partner with SAS and how this has translated into a better product and solution for SAS customers.
The most important takeaway is to consider idea sharing, problem solving, and partnership beneficial for both short-term tactical needs, and long-term strategic solutions. I have provided feedback and worked with developers to find ways around seemingly immovable barriers. I have also provided feedback to make the product align to business needs more effectively. There isn’t a single problem that has been solved in an innovative way, there are groups of problems solved using out-of-the-box thinking.
How may we capture point-in-time information that comes from a lifetime loss execution construct? The CECL content package does not natively support point-in-time output. Since CECL is a lifetime loss construct, it does make sense that only total Expected Credit Loss (ECL) would be calculated and retained. Yet, inevitably senior leadership will want to know the timing of these losses. Further, model performance monitoring requires, and regulators will often ask for this information to ensure losses are not out of sync with Model Development intent. There are considerably more reasons for this, but I provided these points in summary. This required innovative thinking and strong collaboration as it needed to function properly across all products within the risk solution. Working with ways to specialize output variable definitions and allowing each output variable to be recorded for each simulation horizon, we achieved point-in-time capture.
How are we able to capture macroeconomic information as evidence of proper execution? With CECL it is important to understand the scenarios being used and the weights associated with those scenarios if a scenario probability-weighted approach is leveraged. As such, I explained that there needed to be a way to evidence which scenarios and associated weights we assigned. Further, evidence of which modeling system, model group map(s), and associated model(s) is critical. Working in partnership with SAS, evidence for both were added to the MIP execution summary. This drastically simplified the evidence process and allowed for streamlined audit review. This required approaching the problem from a different direction. Looking at macros that call results to capture cycle information and pass this along to the execution summary.
How can we make overlays and adjustments easier in a cycle? In Stress Testing, the ability to add overlays is very important. While the solution provides a path by which the user may make adjustments, specifically using rules in the Risk Governance Framework (RGF), we worked with SAS to develop a step in the workflow which allows the users to enter the overlay amounts in line with the current cycle, no longer requiring the user to leave the cycle all while maintaining the inherent functionality.
How can we reduce human error and be more efficient with complicated model executions? Time is also a factor, so efficiency gains are sought for any part of the process. We worked to solve an automation problem by creating batch script which registers scenarios automatically and executes the credit risk analysis (the quantitative model execution) using batch processing. This allows the user to start the day with results from overnight execution and significantly reduces the risk of incorrect entry.
There are many more examples of problem solving, collaboration, and effective feedback, but the most important point and the best form of a compliment is that many of the problem-solving ideas have been added by SAS in later versions of the solution and products. Further, I have worked with SAS across multiple groups, including Professional Services, Sales, R&D, and Operations.
Thank you for your time and consideration.
How did you use SAS to solve that business problem? What products did you use and how did you use them?
Working in a collaborative way and changing the view on problems. Switch from how to we get around a blocker, to how can we make this work. It sounds like the same, but it's not. The first involves focusing on what is preventing the desired outcome, the second is ignoring these preventions and focus on the problem one is trying to solve. Work backwards from the desired end-state and navigate challenges this way, rather than working forward.
What were the results or outcomes?
New additional features have been added out-of-the box as part of subsequent updates for both the Risk Stratum Core and the different Content Packages.
Why is this approach innovative?
Multiple, see answer to the first question.
