Felipe Millon is the senior manager for housing industry within Amazon Web Services worldwide public sector financial services business.
Across the federal government, cloud computing is giving agencies the ability to experiment with new approaches and processes to accomplish their missions more effectively. These new approaches are opening doors that agencies didn’t foresee, offering them wider applications and greater benefits than agency leaders often first envision.
We’re also seeing these results play out in U.S. housing finance. One of the best examples we’ve seen is with Fannie Mae, a mission-focused organization that enables access to mortgage financing. Fannie Mae moved to the cloud with an enterprise-level focus on gaining greater agility. However, their cloud journey has led to a wave of broader IT experimentation.
Part of this IT experimentation has involved machine learning, for example. As Fannie Mae Vice President Subra Seshadri shared at AWS’s most recent re:Invent conference, Fannie Mae sought to identify and predict which loans in its vast portfolio were most likely to slip into “trouble” within the first two years of acquisition. Because Fannie Mae’s mission is to provide stable sources of liquidity to support housing — by buying home loans, packaging them as mortgage-backed securities, and guaranteeing them — the ability to predict which loans might default is critical.
Fannie Mae had already begun leveraging AWS for a variety of storage and computing services. But as officials became more familiar with AWS’s capabilities, Seshadri’s team began testing AWS’s fully managed machine learning services — using Amazon SageMaker, AWS Lambda, Amazon DynamoDB and Amazon S3 — to build and train machine learning models and deploy them into a production-ready hosted environment. By testing a rapid succession of models inexpensively, and machine learning, Fannie Mae was able to boost its predictive “catch rate” on multifamily loan applications likely to default from 3.5% to 48%, according to Seshadri. That experiment helped Fannie Mae plan for early risk mitigation on higher risk loans. It also opened the door to exploring other cost savings opportunities at Fannie Mae and accelerated the pace of digital learning and transformation more broadly.
Fannie Mae’s focus on innovation also illustrates how cloud services can enable “two-way door” decisions. Traditionally, when agencies make IT investment decisions, they almost always have to live with those decisions for a long time. That’s a “one-way door” decision. A two-way door decision lets you make an IT investment, and if you aren’t getting the results you want, you can easily reverse that decision.
That’s a powerful concept in IT. The cloud is turning a lot of what once were one-way door decisions — for investing in infrastructure, applications and IT services — into two-way door decisions. Now with the cloud, a team can simply spin up a virtual server, try out a variety of applications or data analytics capabilities, and see what works. If they like what happens, they can scale it up quickly. If they don’t, they can easily shut it down, without the usual upfront capital costs and long-term expenses.
By having much greater freedom to experiment, organizations come to a greater understanding of how to use, benefit from, and become more productive by leveraging the cloud.
Organizations also gain something even more important: By removing what we call the “undifferentiated heavy lifting” — the basic IT hardware and software that supports, but rarely distinguishes, your organization — they can focus more resources on delivering their missions.
Reducing that heavy lift has become even more important in federal and public sector organizations, where they face a steeper climb trying to assemble and retain the right IT talent. At AWS, we’re helping to make that lift even easier — not only by providing the latest turnkey tools and engineering support, but also by helping organizations train and certify their IT staffs.
We’re excited to see how these kinds of experiments are unlocking innovation across the U.S. housing finance system and throughout the federal government. The results show that, with the right leadership and empowerment, the cloud is enabling organizations to establish big goals, experiment, and push through once-intractable barriers.