How Policymakers Fight a Losing Battle With Models
This article appears in the April 2023 issue of The American Prospect magazine. Subscribe here.
If you’re wondering why the U.S. has failed so miserably in developing a workable child care and early-childhood education system, consider the role of economic modeling.
In 2021, when the Congressional Budget Office (CBO) released its much-anticipated score for the cost of the child care provisions in the Build Back Better Act, it produced one headline number: $381.5 billion. This was what CBO estimated as the amount of money the government would lay out for child care.
But that budget score badly missed the mark on the net cost of the program. It did not account for any of the savings predicted by reams of academic research on the long-term economic benefits of child care. Nothing about how kids with high-quality early care do better in school, stay out of trouble, and have higher lifetime earnings. Nothing about the increased tax revenues generated by mamas and daddies who could now work full-time. Nothing about the mountains of data that show that when mothers are held out of the workforce in their early years, their lifetime earnings and even their security in retirement are seriously undercut—something universal child care could reverse. And nothing about the impact of higher wages for child care workers—wages that would mean many of those workers would be paying more taxes and wouldn’t need SNAP, Medicaid, housing supplements, and other help offered to the lowest-paid people in the country. In other words, according to CBO, investing in our children and filling a wheelbarrow with $381.5 billion in cash (a big wheelbarrow) and setting it on fire would have exactly the same impact on our national budget and our nation.
To every CEO of a Fortune 500 company or owner of a small neighborhood restaurant, budget scoring like this must sound like a crazy way of doing business. After all, investments don’t just have costs—they also have benefits. That’s why companies invest in things like building factories, converting to green energy, or offering employee benefits, even if they have to book a big cost up front. Those corporate executives don’t take on big-ticket projects out of the goodness of their hearts; they take them on because they want to boost profits, retain workers, and improve the company’s long-term outlook.
Budget rules, by contrast, tilt against investing in people. And there’s a reason for that. Decades ago, Congress decided that CBO cannot account for the indirect or secondary effects a policy change may have on other parts of the budget. Research shows, for example, that federal spending on things like safe housing and nutrition assistance for babies makes people healthier and reduces total health costs. But because of the rules Congress set, CBO cost estimates for these programs cannot assume taxpayers would save any money on health insurance costs or that taxpayers would spend less on Medicaid. Meals on Wheels helps seniors stay out of much more costly nursing homes and saves Medicare and Medicaid billions of dollars, but the federal government says it is nothing but an expense. Beefing up IRS enforcement, as Democrats did in the Inflation Reduction Act, would mean fewer tax cheats and more revenue. But according to official CBO scoring, more money for the IRS is mostly another expense that adds to the deficit.
Bad budget modeling, and how members of Congress respond to it, also distorts the way we design policies.
These and other self-imposed rules structurally bias the policymaking process away from making commonsense investments that meet families’ needs.
Bad budget modeling, and how members of Congress respond to it, also distorts the way we design policies. Consider again our country’s need for child care. The U.S. is 33rd out of the 37 richest nations in terms of what we spend on child care, and millions of parents—mostly mamas—are kept out of the workforce because they can’t find safe, affordable child care. The pandemic drove this crisis into the open, fueling a national outcry over the sorry state of care for our youngest children.
When I was invited to deliver one of the keynote speeches at the all-remote 2020 Democratic convention, I spoke from a closed child care center in Springfield, Massachusetts—standing amid the blocks, tiny chairs, and individual cubbies in the room for three-year-olds. As more people rallied behind the need for universal child care, and as Democrats won both the White House and Congress, I believed this was our moment.
But as I assembled a new, comprehensive bill, the first question I got was the dreaded “How does it score?” The answer hobbled the process from the start. Instead of fighting for good policy, Democrats arbitrarily decided that the child care provisions in the bill would need to cost less than $400 billion. Universal care costs a lot more than that.
So the bill that ultimately moved forward was not based on how much money it would take to make certain every child had access to care. Instead, it was loaded with ways to game the policy design so the CBO score would meet the $400 billion threshold. The bill cut out millions of families that needed care, delayed implementation for years, and let states opt out. Bad budget modeling meant that these decisions weren’t driven by what would maximize our children’s well-being or our nation’s long-term growth. Instead, decisions were driven by the political imperative to produce a smaller score, regardless of what it meant for the workability of the proposed program.
The only number used to evaluate a child care program was the direct outlay for care. The compromises made to hit a politically palatable CBO number raised questions about who would and wouldn’t benefit from a compromise program, draining away support. As one bill after another passed the 2021-2022 Congress, child care was left behind.
OUR CURRENT BUDGET MODELS don’t create random errors. They don’t sometimes overstate and sometimes understate costs. They create systematic errors, making many investments look far more expensive than they are. They lead to the routine underfunding of critical programs and enforcement activities, and they distort the policymaking process from start to finish.
For those who want to shrink the government to a size that can be drowned in a bathtub, the current budget-scoring model works great. But for those who live in the real world and want a country in which all our people have a chance to thrive, bad budget models are choking us.
Reforming these economic models would not be easy. CBO cost estimates generally exclude the potential macroeconomic effects of a proposed policy precisely because, as they explain it, they have too few analysts to crunch the numbers. Worse, say former CBO scorers, the “estimates of macroeconomic effects are highly uncertain.” Translation: It’s hard.
Yes, estimating the costs and benefits of major investments decades out is hard—really hard. Figuring out the right model and the right assumptions is tricky and uncertain, and real life can prove our best estimates wrong. Politics can weave its way into judgment calls. Data are imperfect. But “hard” is no excuse for not trying.
Congress has a responsibility to maximize our people’s long-term prosperity, and we need economic models that stand a better chance of doing that.
We should start by changing the rules on modeling so that both costs and benefits are accounted for. And because this is hard, we should ensure that the agencies we rely on for modeling have the resources they need to provide a solid understanding of both the likely costs and the likely benefits of any given policy.
We should also ensure that our modelers, who make countless judgment calls in doing their work, reflect a real diversity of perspectives. Consider CBO’s Panel of Economic Advisers. CBO relies on these experts, “selected to represent a variety of perspectives,” to help calibrate their models and gut-check their assumptions. But look at the team: Of the 22 people on last year’s panel, 20 had doctorates in economics; 11 of those 20 went to the same three Ph.D. programs. Recognizing that economic modeling is highly uncertain and highly dependent upon assumptions means that we should strive for diversity among our modelers and be open to different kinds of models when making decisions. CBO shouldn’t be the only game in town.
We should also build accountability into the modeling system. Instead of scoring, voting, and moving on, we should assign independent outside teams to collect data on programs that have been adopted to see how far wide of the mark past modeling turned out to be. That information would arm us to improve modeling over time.
Finally, policymakers need to remember that modeling the costs and benefits of major public policies isn’t just about numbers—it’s also about our values. Yes, we need better data and better models, but we also can’t be afraid to make the case for bold investments, even when the (accurate) price tag is high. I believe that with accurate modeling, high-quality child care pretty much pays for itself. But even if not, such care helps us create an America in which everyone has opportunities—and “everyone” includes mamas. An investment in care is also an investment in care workers, treating them with respect for the hard work they do. In other words, just as a CEO would make the case to her board for taking a bet on a big new project, Congress shouldn’t shy away from making the investments the American people and our economy need.
Elizabeth Warren is the senior Democratic senator from Massachusetts.
TAP (The American Prospect) depends on your support. We've have said it before: The greatest threat to democracy from the media isn’t disinformation, it’s the paywall. When you support The American Prospect, you’re supporting fellow readers who aren’t able to give, and countering the class system for information. Please, become a member, or make a one-time donation, today. Thank you!