The Most Important Economic Policy Model Nobody Understands

The economic models that can make or break legislation in the U.S. are fundamentally flawed.

Alberto Miranda for Noema Magazine

Mark Paul is an assistant professor of economics and environmental studies at New College of Florida. He is a 2021-22 Berggruen Institute Fellow.

You might think that the Build Back Better Act died in Congress because the numbers didn’t add up. The Congressional Budget Office (CBO) crunched the numbers and concluded that the plan would increase the deficit, and news coverage ran with that headline. This was enough to provide cover for swing votes in Congress, and Sen. Joe Manchin, in an extraordinary interview on Fox News Sunday, stated he would vote “No,” citing the true cost of the bill as determined by — you guessed it — the CBO.

But the numbers lawmakers rely on to gauge the potential economic impact of legislation are wrong — and systemically so. The CBO’s model is built on a whole host of false and outdated assumptions, including that public investment is less productive than private investment, that taxes — especially progressive ones — tend to have a negative effect on economic growth and that externalities like climate change don’t factor into the economic impact of legislation. This flawed model has routinely sunk attempts to invest in the economy and the American people, rein in the wealthy and eradicate poverty.

The CBO has the power to make or break legislation. As Sen. Chuck Grassley put it back in 2006, “CBO is God around here, because policy lives and dies by CBO’s word.”

This isn’t just hyperbole. Take, for example, the federally mandated minimum wage. Today, it is a miserable $7.25, less than half what it would have been if it had kept pace with productivity gains (which is how it used to be determined). It’s even lower for tipped employees, who in many states get just $2.13 per hour from employers. Increasing the minimum wage to $15 would effectively end poverty-level wages overnight by boosting wages for 40 million workers, most of whom are desperately struggling to make ends meet. The CBO, however, wrongly reports that raising the minimum wage will lift 900,000 people out of poverty but put 1.4 million people out of work and prove costly for the economy overall. As a result, the legislation to raise the minimum wage has yet to receive a vote in Congress; instead, it’s stuck in purgatory.

Why does the CBO’s model — which almost nobody understands — have so much power over the economy?

No other country has a budget agency with the political clout of the CBO. Before 1974, major legislation — including all New Deal and Great Society programs — passed without any official score from the CBO. They didn’t even exist yet. But Congress was frustrated with the Nixon administration for impounding funds that it had already appropriated in instances where the administration disagreed with Congress — and for essentially cooking the budget books by directing the Office of Management and Budget to publish biased, partisan economic reports in his favor. Congress wanted a neutral party to evaluate the economic impacts of legislation, one without partisan affiliation and with credibility on both sides of the aisle. 

“This flawed model has routinely sunk attempts to invest in the economy and the American people, rein in the wealthy and eradicate poverty.”

In comes the CBO, whose job is straightforward: they are tasked with crunching the budget numbers for members of Congress on proposed legislation. A member wants money to fund a new program to address the issue of homelessness. But how much will it cost? Will this increase or decrease GDP at the end of the day? The CBO’s job is to estimate the costs and economic impact of the bill. However — and this is very important — the CBO doesn’t make policy recommendations. As Alice Rivlin, the first director of the CBO explained in an interview, one of the most important decisions early on in the creation of the group was to “protect the nonpartisanship of the agency” by ensuring the “CBO would not make [policy] recommendations.”

This laid the groundwork for CBO’s sweeping influence and durability on the Hill. It’s seen as a credible and reputable arbiter. It’s the scorekeeper, with no political affiliation. And indeed, it has drawn fierce criticism from both Democratic and Republican administrations when its score doesn’t support the political priorities of the sitting President — the Trump administration, for instance, launched an attempt to undermine the CBO’s credibility. But Congress respects the office for its supposed neutrality. And legislators are willing to go to great lengths to get a “favorable” CBO score by tweaking legislation to game the system. 

The CBO isn’t the only one providing these headline-grabbing scores that can decide the fate of legislation. Lawmakers also turn to the Penn Wharton Budget Model, a Wall Street-funded private entity that tends to use even more conservative assumptions than the CBO, for an alternative estimate, giving this non-government entity a great deal of power to influence legislation with its model. While Penn Wharton claims to be nonpartisan and transparent, I have dug at length into their limited public documentation and can say that they’re anything but.

“All models are wrong,” the great British statistician George Box wrote in 1987, “the practical question is how wrong do they have to be to not be useful.” Unfortunately, the macroeconomic policy models used by the CBO and Penn Wharton happen to be very wrong and thus, perhaps, not of much use.

Take the models’ treatment of public investments. Time and time again, the CBO and Penn Wharton suggest public investment is a bad idea for the economy, often shrinking GDP and thus leaving workers worse off and the government further in debt (which both organizations view as categorically undesirable). At best — as Penn Wharton found in its analysis of the bipartisan infrastructure bill — they’ll say the legislation will increase government debt but be a wash in terms of economic growth. But how can this be? The country desperately needs investment in its ailing infrastructure. In fact, the American Society of Civil Engineers gives the nation’s infrastructure a C- and argues that the country’s decrepit infrastructure costs the average American household $3,300 a year.

“The true costs of climate change just don’t exist in the models and therefore policies aimed at mitigating it inevitably seem costly.”

Despite this, the scorekeepers assume public investment is far less productive than private investment. But take their word for it, not mine: “CBO further estimates that productive federal investment has an average annual rate of return of about 5%, or half of the agency’s estimate of the average rate of return on private investment.”

This estimate, however, isn’t grounded in empirical literature; in fact, empirical evidence shows that public investment often has a higher economic rate of return than private investment. But it’s not a firm either-or proposition when it comes to public investment or private investment; the vast majority of the time we do not need to choose between public and private investment, nor is that the only evaluation we should use when considering legislation.

What’s more, the models painstakingly detail the “costs” of public investment while largely failing to account for its benefits. Legislation designed to address the climate crisis is a case in point. Both the CBO and Penn Wharton models fail to account for any benefits from averting air pollution or reducing carbon emissions — and thus averting dire levels of global warming. That’s a pretty huge shortcoming when, for example, climate fueled disasters are wiping out economic growth. (In 2018, more than 60% of the year’s economic growth was offset by hurricanes, fires and droughts and other severe weather.) And it’s not just money that we’re losing to climate change — every year an estimated 250,000 Americans die premature deaths due to air pollution that is primarily the result of combusting fossil fuels. Yet, in the “scores” of record, these effects are ignored. The true costs of climate change just don’t exist in the models and therefore policies aimed at mitigating it inevitably seem costly.

The models also misrepresent the relationship between public and private investment. For example, both CBO and Penn Wharton assume public spending necessarily reduces private (i.e., more productive) spending — what economists call the “crowding-out effect” of public investment. But empirical literature shows that, if the economy is operating below capacity — meaning there is unemployment and idle resources, which has almost always been the case in recent years — public spending will create more opportunities for private spending. Public spending means more jobs. More jobs mean workers with cash in their hands. Workers with cash tend to spend it, driving up demand and “crowding in” — encouraging more — private investment.

“The models painstakingly detail the ‘costs’ of public investment while largely failing to account for its benefits.”

We can clearly see that the CBO’s past predictions about the negative consequences of public investment have not come true. Despite consistent projections from the CBO, public spending — and an increase in government debt — has not led to higher interest rates (which would make borrowing more costly and “crowd out” investment). Interest rates remain well below projections as investors continue to buy up safe government assets. (In fact, it was the shortage of safe government debt that helped fuel the housing crisis and the Global Financial Crisis.) Furthermore, modelers fail to acknowledge that interest rates are not purely market determined, but rather policy variables that can be managed by the Federal Reserve. In other words, there is not some fixed pot of money that public and private actors are constantly competing for — just look at how Congress appropriates gobs of money for “defense.”

The results these models produce are not objective reflections of reality, but the product of ideological assumptions. Among the most glaring examples is the assumption that public spending always takes away from private spending and drives up borrowing costs. They reproduce a textbook neoliberal view of the economy, where the rich are the job creators, inequality doesn’t matter and the government is simply an incapable actor.

The models make similar assumptions when it comes to taxes. The CBO found that typical taxes the government may impose as a part of social spending legislation will have significant negative effects on economic growth, and in turn, job creation. (It’s important here to note that increases in public spending need not entail a commensurate increase in taxes.) Taxes on corporations and high-income people (i.e., progressive taxation) are the worst offenders in their models. But evidence increasingly shows that not only are the negative growth effects from such taxes overstated, but these taxes may actually help increase growth, as firms with excessive market power are taxed.

The models also do not account for the power that individual companies have to set prices. Big firms, as we’ve seen over and over again in recent years, have outsized power in markets. Amazon, for example, effectively decided to take over the diaper market in 2009. After undercutting competitors by up to 30%, it changed course. Once Amazon became the dominant diaper retailer, it started to jack up prices, leaving consumers with little to no choice but to pay up.

Big corporations often have the power to dominate prices for workers, too. When there are few employers in town — or, at least, few in the line of work that employees undertake — then workers’ options become limited. As a result, firms have what is known as monopsony power — the ability to use their size to undercut workers’ wages and well-being. Simply accounting for market power in the models would be a sizable step in the right direction.

Further, the models tell us little about how the poor fare vs. the rich, when it is possible for models to break down economic impacts on different demographics. And neither of these organizations publish in a transparent way the data and full assumptions that go into their models, denying policymakers and the public the opportunity to understand their results. (Beggaring belief, among the Penn Wharton model’s many fallacies is the assumption that undocumented immigrants are inherently less productive than other workers.)

“It’s simply not true that many of these ‘post-neoliberal’ proposals are bad for the economy, yet our models are holding us hostage.”

For some, reform doesn’t go far enough. Instead, they argue the CBO should be dethroned, or simply abolished. And that may very well be the best path. But with some fundamental adjustments, the models deployed by the CBO and Penn Wharton could in the least inform lawmakers and the general public accurately. Those adjustments begin with the models taking into the account the past two decades of economic research. Gone are the days when phenomena like market power, distributional concerns, productive public investment, and externalities (such as climate change) could simply be treated like afterthoughts with little to no economic implications. Times have changed and so too must the models.

Simply assuming that public investment is at least as productive as private investment would drastically change the calculations. All of a sudden, public investments wouldn’t be thought of as a wasteful endeavor; instead, they would be treated as they are: crucial productive investments that underpin the entire economy and our collective well-being. After all, how productive would we be without roads and public transportation, without public education?

And Congress has the power to force these kinds of changes. As Philip Rocco, a political scientist who studies the CBO, has argued: “Never forget that the CBO is a legislative support agency created, funded, and directed by Congress.” If Congress recognizes that the modeling choices made by the CBO aren’t indeed reflecting the real world, then Congress “can alter them [the models and assumptions used by CBO] when they want to.”

That’s precisely what some members of Congress want to do. The Carbon Pollution Transparency Act of 2014, introduced by Sen. Bernie Sanders, would require the CBO to calculate a carbon score for legislation; in 2021, Sens. Elizabeth Warren and Michael Bennet introduced the FAIR Scoring Act, which would mandate the CBO measure the impact legislation will have for low-income families and communities of color. Even Republicans are calling for changes. Last year, led by Sen. Mike Lee of Utah, Republicans introduced the CBO Show Your Work Act, which demands that the CBO provide all of the data it uses publicly and allow for its analysis to be replicated by outside groups. (To be clear, I have no illusions that Democrats and Republicans will agree on how to change the models.)

It’s simply not true that many of these “post-neoliberal” proposals — to address poverty, support workers, invest in housing and children, and usher in the 21st-century green economy we desperately need — are bad for the economy, yet our models are holding us hostage. The models may say so, but the models are dead wrong. It’s time for new models that better reflect our reality and account for the very real dangers, from the climate crisis to unchecked corporate power, that threaten our economy. If we don’t recalibrate the way we calculate the costs of potential legislation, it will cost us all.