bannerimage
Ginnie In Brief
 
Share
* To
* From
Message
URL
https://www.ginniemae.gov/newsroom/GinnieInBrief/pages/Post.aspx?PostID=55
Print Friendly

​​​​​​​​​​​​​​​​​​

ARTIFICIAL INTELLIGENCE AT GINNIE MAE
by Ginnie Mae | 5/17/2021

Artificial intelligence (AI) — or the use of computer algorithms to perform tasks such as learning and problem solving that have traditionally required human intelligence — is no longer the stuff of science fiction. From supporting customer service requests via chatbots to enabling virtual assistants to respond to voice commands, AI is transforming virtually every element of how we live, work and interact with one another, including the home lending industry.

More than half of senior mortgage industry executives in a Forbes study agreed that AI is revolutionizing the business. AI reaches into all aspects of the lending process, from loan application and origination to underwriting and servicing, and it may disrupt aspects of these processes that have been in place for years. AI is enabling new players to enter the market, sparking greater competition, and revealing a potential need for new regulations. In 2017, over half of financial services firms had made investments in AI technology, and analysts project that most mortgage lenders will use AI to transform some parts of their operations by the end of 2021.

Organizations in the home lending space are already leveraging AI and machine learning in a variety of ways to speed up processing and automate tasks that used to be manual and time intensive.

For example, Freddie Mac is integrating AI and machine learning into their Automated Collateral Evaluation to determine if a mortgage is eligible for sale, bypassing traditional requirements for an appraisal report.

At Ginnie Mae, our forays into AI and cognitive technologies have largely centered on Robotic Process Automation (RPA). One bot that we’ve put in place, called DABO, streamlines the previously manual process of determining adjustable mortgage rates. Another bot tackles the task of managing and reporting commitment authority reconciliation data to the general ledger, a task that previously took three hours and at least three employees to accomplish. With RPA, Ginnie Mae was able to reduce that time to one hour for a single staff person. By automating these processes, Ginnie Mae has reduced the time employees spend performing manual tasks and has also significantly reduced errors, freeing up employees to focus on other, more valuable activities and projects.

Over the last three years, Ginnie Mae’s Office of Enterprise Risk (OER) has launched a series of machine learning and AI model pilots, utilizing different approaches to explore new ways of measuring and analyzing data. One AI algorithm similar to the one used in genomic sciences is being deployed to reduce the probability of false negatives and false positives when identifying Issuers that may pose enhanced risk to the program but may slip through the cracks of traditional methods of risk identification. We plan to use this algorithm in our issuer risk management processes by fall 2021.

OER has also explored cutting-edge graph technology to create a digital replica of the mortgage financing markets — mapping out the complex network of issuers, their creditors, parental institutions, and other players in the mortgage finance ecosystem. This simulation can identify cascading effects on the greater ecosystem, drive proactive policy analysis, and perform Black Swan simulations. In another nod towards efficiency, OER is putting in place a machine learning tool to automatically review Issuers’ credit by translating financial metrics into qualitative assessments. The tool, which also will be live in fall 2021, will reduce the workload of credit analysts significantly.

AI can also automatically flag missing or inconsistent loan data. Usually, it is the job of the underwriter to verify the data, and many underwriters are still manually inputting data by transcribing it from paper documents to spreadsheets. With AI, underwriters can focus on dealing with automatically flagged errors and exceptions, rather than searching for the error with their naked eyes.

The bottom line with AI is that it soon will be ubiquitous throughout our industry and embedded more deeply into Ginnie Mae’s processes, becoming an ongoing force for broader access to low-cost mortgage financing — just what Ginnie Mae has been working to do since our founding more than 50 years ago.

​​​​
Ginnie in Brief Contributors
Laticia Jefferson
Richard Perrelli
Barbara Cooper-Jones
Regina Chase
Angel Hernandez
Alven Lam
Eric Blankenstein
Seth D. Appleton
Omar Bouaichi
John T. Daugherty
John F. Getchis
Roy Hormuth
Tamara Togans
Maren Kasper
Gregory A. Keith
Michael Drayne
Ginnie Mae
Michael R. Bright
Latest Posts
Ginnie Mae Digital Collateral Program Exceeds $2 Billion of Issuance
Ginnie Mae MBS Through an ESG Disclosure Lens
The Ginnie Mae Innovation Lab Embarks on Blockchain
Enhancing MBS Disclosure Will Deepen Demand for Ginnie Mae MBS
Last Modified: 3/23/2021 10:16 AM