Ginnie Mae’s data processing operations manage millions of data records each month, including the information that enables the flow of billions of dollars in principal and interest payments to investors around the world. To keep this enormous task running as smoothly and efficiently as possible, Ginnie Mae is investing in the development of artificial intelligence (AI) capabilities led by teams within the agency’s Office of Securities Operations and Office of Enterprise Data and Technology Solutions.
These efforts align with the Presidential order on AI from earlier this year, which serves as the basis for a whole-of-government strategy to tap private sector innovation in support of government program excellence.
AI encompasses a range of applications, processes and technologies. At Ginnie Mae, robotic process automation (RPA) is the first type of AI that we have deployed. Despite “robotic” as part of its name, RPA does not actually involve robots or other physical manipulative assets; RPA uses software to replicate repetitive human tasks such as the collection and analysis of data. (see Figure 1)
In the second half of 2019, Ginnie Mae completed two RPA projects, or bots. The first is designed to collect and organize data related to the London Inter-bank Offered Rate (LIBOR), the principal index for the majority of adjustable rate mortgage-backed securities. The second bot assists staff within the agency’s Chief Financial Officer division to manage and report key information into the Ginnie Mae general ledger. Both bots free our staff to work on more value-added tasks, increasing the overall efficiency of the agency.
Although RPA is an excellent tool for increasing the efficiency of repeatable processes, it has also been important to recognize that not every repeatable process is the same. For example, research shows that RPA is best applied to processes that use structured and accessible data sources, such as the publicly available data on LIBOR or in-house financial statement data. RPA can also be used with unstructured data with a clear rules-based protocol for data manipulation. However, any break from the rules could cause an exception, or breakdown, that would require staff intervention, ultimately rendering the process ineffective.
As Ginnie Mae moves forward with its strategic and technology modernization plans, the agency intends to expand the number of RPA processes deployed and implement more sophisticated AI functions, such as machine learning, where appropriate.
Each phase of our modernization strategy will be governed by what is fiscally sound and secure. Our plan is to leverage capabilities of the private sector, as defined in the aforementioned Presidential order, as well as adopt or develop capabilities in partnership with leading-edge firms that advance our technological infrastructure. This strategy will effectively meet the changing needs of a sophisticated and nimble mortgage finance market, while keeping true to our mission to protect taxpayers and provide a robust secondary mortgage market for government home loans.