Why the rollout of the $2 trillion CARES Act was a colossal mess

Desperate citizens haven’t gotten their stimulus checks. The Payroll Protection Program has run out of money. According to two former federal employees, the CARES Act was doomed from the start.

Why the rollout of the $2 trillion CARES Act was a colossal mess
[Source Images: nidwlw/iStock, xxcheng/iStock]

Imagine for a moment that you are a member of Congress working on the bill that is going to help save the country from economic devastation. Unemployment is at a record high, business of all sizes are hemorrhaging money, the general populace is simultaneously terrified and bored. Where do you start? As researchers, product designers, and product managers well know, when designing anything, you start with the end users. What do people need? What’s the best way to meet those needs? How will people access whatever the end product is?


If you are a member of Congress, you didn’t start with any of these questions, because this is not how the standard policy design process works. Instead, you looked at what has and hasn’t worked before. What should the Fed do? What about a cash infusion into failing industries? What about sending people checks? A shrinking economy needs money to flow, and the best way to do that is to get dollars in the pockets of people who need them, and will spend them right away. This economic analysis is important, deep work, but it also isn’t nearly enough. Nowhere in the process is there a step in which policymakers think about delivery—how do people get those dollars?—and how that might impact policy design. Which is how we ended up with a $2 trillion stimulus package that doesn’t factor in what people actually need or how they might have those needs met. As a result, a lot of that money will never reach the people it is intended to help.

A good idea poorly executed

The CARES Act—Coronavirus Aid, Relief, and Economic Security Act—allocates funds for large business, small businesses, airlines, and individuals. Individuals can request $1,200, more if they have children under 18. The bill also increases unemployment benefits—providing an additional $600 on top of what people can get from their state. But a complex web of bureaucratic requirements means that lots of people won’t be getting what they need anytime soon, and maybe not at all.

As of this writing, only half of an anticipated 150 million payments have made it into people’s accounts. There are myriad reports of people receiving “status not available” messages from the IRS portal when trying to see when they might get their checks. Others weren’t even able to log in. The Payroll Protection Program (PPP), part of CARES aimed at helping small businesses, just announced that it has run out of money and cannot accept any more applications. And as if the process weren’t cumbersome enough, the checks will also be delayed in order to emblazon President Donald Trump’s name on them, as though they were a hotel or a golf course. By contrast, citizens of other countries around the world reported receiving their checks within hours of applying.

“For us, it hasn’t really solved anything”

Margaret Coleman runs the T.W. Wood Gallery, an arts nonprofit in Montpelier, Vermont, that earns its revenue from after-school art programs and donations. Those two sources evaporated as a result of school closures and the economic downturn. She had to lay off her entire staff over the past few weeks. She cut her own hours down to 10 hours a week, which she says she spends exclusively applying for grants. (With two young children at home, Coleman doesn’t have oodles of time to work, so 10 hours is about all she can manage to eke out.) Coleman says the amount her organization needs to make it through the next few months is about $10,000, but there is nowhere to get that amount of money. “Compared to other organizations, that amount is minimal,” she says. “But it’s not there.”

Her nonprofit should be eligible for $18,000 through the PPP, but because that money can only be applied to payroll, she would need to rehire the employees she laid off, at a rate several hundred dollars less than they are able to earn through unemployment insurance. So instead she is looking to hire new people at $16 an hour for the two-month period covered by PPP. At which point schools won’t have restarted, the economy will likely still be in tatters, and the money from CARES will be gone.


“For us, it hasn’t really solved anything,” Coleman says. “It’s going to keep us in this weird floating position for two months. But then, when the organization is hopefully able to open back up, we won’t have anything to fall back on.”

[Images: nidwlw/iStock]

User research is missing

What would have helped, Coleman thinks, is putting the money toward the organization’s mortgage. This is exactly the kind of detail that reveals itself in big, bold letters when you conduct user research. But we know that Congress did not speak to end users in designing this bill because as former federal employees—at the Department of Homeland Security and the Obama White House—we know there is no process in place for this to happen.

When there isn’t a time-sensitive need (like a global pandemic), Congress makes an effort to hear from different end users (or in political terms, constituents) by calling experts who represent different points of view. They may summon someone who represents people with disabilities, or they will take calls from constituents, but this process only approximates deep user research. Some groups will not be represented. Additionally, the experts who speak to Congress may have a pretty well-informed sense of what people need in general, but those experts are rarely members of that user group themselves. And when it comes to something like receiving a stimulus check, it is not enough to learn that people need money, you need to go deeper to understand the barriers (fast internet, bank account) that would allow people to access emergency cash. User researchers are often asked if someone can pretend to be a different user group to save research costs. (“I can imagine what it’s like to be homeless,” for example.) But there is no substitute for hearing directly from real users.

The experts do the best they can, but there is no mechanism for sending out a survey or conducting user research and hearing the voices of those who have just lost jobs and don’t know where their next paycheck might come from, who are facing a pandemic without health insurance, or who are under shelter-in-place orders without a home to shelter in. For critical swaths of the public there is no voice, and therefore no ability to truly empathize and understand what help looks like.

As checks have begun to arrive, we can see what people’s most critical needs are by how they are spending the money. One early report shows that the stimulus money is going toward food and gas, though people quoted in the same article say they are using the money for everything from rent to student loans. There is value in understanding how the money is spent, and if the people who crafted the bill had been able to use that data as a starting point—what do people need help with and what does that help look like?—the resulting bill might have looked very different. Maybe a structural change such as raising the minimum wage is the solution. We will never know because we failed to ask.


It is also clear that delivery was not thoroughly thought through, in part because it took two weeks from the time the stimulus bill passed for the IRS to announce that they would be launching a portal for people to get a handle on when their payment would arrive. Government often announces the development of a seemingly simple piece of technology without having plans in place for actually building the thing, for a range of reasons including literal walls or even miles separating the policymakers and the technologists. Which means that a policymaker may announce a thing only to discover that an agency doesn’t actually have the data available to make that thing a reality.

Sometimes the very structure of a bill accidentally leaves out the very neediest. In order to get a stimulus check from the federal government you have to file taxes. But 10 million low-income families don’t earn enough to file taxes. It took weeks and urging for the IRS to build out a site for non-filers. Even people who do file taxes aren’t necessarily covered. One hundred million taxpayers don’t have direct deposit info on file with the IRS, which means that they won’t see money until checks can get printed and mailed, which could take months. As we were writing this, the IRS launched a way for people to provide this information. And for those waiting for their checks to arrive by mail, we’ll have to hope that the Post Office doesn’t go belly up in the interim.

Government ‘can’ get it right

When government digs beyond the surface to truly understand an issue, positive changes happen. In Mobile, Alabama, a team working on reducing the number of blighted properties found multiple unexpected causes, including a small piece of the state constitution that needed amending, and were able to significantly reduce the number of blighted properties in the city. A team in Phoenix chipped away at underlying causes when looking to improve recycling rates and also had success. For governments willing to look at the thing they are trying to solve as one piece of an intricately intertwined system, real, lasting change can happen. Congress has yet to take this approach.

So what should the policy design process have looked like for the stimulus bill? A quick check-in of user needs—what do people most need to pay and when—would have provided real world, immediate user data for policymakers to use as a guide. Similarly, conversations with the state and federal employees who process unemployment benefits would have yielded a gold mine of interesting and relevant information. Did those employees have any concerns as the number of people was set to skyrocket? Could their current process handle the crush of applications? Was there any concern that state unemployment websites might crash under the additional load? Were there any workarounds or outdated policy procedures that could perhaps be removed?

In 2015, when the government prepared for a tsunami of asylum applications from Syrian refugees, the Obama administration and the State Department and Department of Homeland Security made sure the system could handle the new load by reworking and digitizing the application process. Previously applications sat in a pile for as long as eight weeks, waiting for a refugee officer to physically show up and stamp them. By creating a digital stamp that allowed officers to review and stamp applications remotely, the government was able to admit more Syrians in one month than had been admitted in the previous seven months.


A final step for the stimulus bill: Once lawmakers had designed solutions based on what people really needed, they should’ve spent time thinking about how to deliver those solutions, and how that delivery might affect the policy itself. These are the annoying details of who will do what when and what. This is also very often the place where government fails. Sometimes this means people are filling out forms in one location and presenting identifying documentation in another. Sometimes it means the only place to submit a form is at some out-of-the-way location during business hours. It is why people might have to take off from work to apply for food assistance or take three trains and a bus with young children in tow to be placed in a new family shelter. It is the reason why interacting with government often means long lines and long hold times and endless waits.

Policymakers could have even run a quick one- or two-day test to see if the proposed solution had the desired effect. Sometimes there isn’t time to do this critical step—in our current environment, where people are out of work and need to make rent and buy groceries now, running a test may not have been optimal. But even when the timeline is short, government can send up a few test balloons.

This isn’t any one person’s or agency’s fault. Our government was built in a different time for a different time. The architecture of the IRS, the agency charged with delivering today’s stimulus checks, started over 200 years ago. Keeping up with the digital age is challenging for all kinds of institutions, especially for cash-strapped state and federal agencies. Some of the smartest, most right-headed people we know work at all levels of government. But it is at these most critical moments, when we need government the most, that the cracks in how government functions are split wide open for all to see. No one knew until last week that most government technology is written in COBOL, a language that is a punchline for anyone who works in tech. Well, no one except those of us who have worked in government, and who have been clamoring for years about how dangerous it is that government’s most critical services run on the equivalent of a rusty telegraph machine. Now the world knows.

In the middle of a crisis, people need a functional government today more than ever. Congress can’t wait until there is a nice chunk of downtime to upgrade the policy design process. This crisis demands new ways of thinking and acting. Starting today.

Hana Schank and Tara McGuinness are working on a book about public interest technology to be published next year by Princeton University Press. Schank is the director of strategy for public interest technology at New America. She previously ran a digital service team working with Customs and Border Protection and TSA at the Department of Homeland Security. Follow her on Twitter. McGuinness runs a lab at New America focused on designing policies for family economic security and well-being. She was a senior advisor to President Obama and oversaw the sign-up campaign to get 7 million people Obamacare. She teaches public policy in Georgetown’s McCourt School. Follow her on Twitter.