During last year’s U.S. presidential debate, members of the Wharton community may have heard a familiar name: Both candidates mentioned the School, referencing its research on their economic plans as a means to bolster their credibility. The name drop left Penn- and non-Penn-affiliated viewers alike wondering: Who are the brains behind this data-crunching analysis that has earned the respect of the most powerful decision-makers in the country?
Combining Wharton’s economic expertise with a dash of political know-how, Penn Wharton Budget Model has built a name for itself since its birth a decade ago due to the strength of its small yet well-rounded team. Research by the initiative, which was founded by faculty director Kent Smetters, has been picked up by major news outlets, including The Hill, Fox News, the Washington Post, and the New York Times.
As the School began to exhale for the summer, Smetters and other PWBM leaders reflected on the unit’s early days and offered a behind-the-scenes glimpse into how it meets the challenges of the current moment. The PWBM team shares a sensible headquarters in Philadelphia a couple of blocks from Penn’s campus, complete with graphs scribbled on whiteboards and windows shaded with Venetian blinds. But as Smetters recounts its origin, he goes back to the 1990s in Washington, D.C., when he was a newly minted PhD with a background in computer science, looking to take an alternate route to the traditional academic path. He landed at the Congressional Budget Office, which was established by the government in 1974 to provide cost estimates and technical assistance to senators and representatives before their bills hit the chamber floor.
PWBM Faculty director Kent Smetters
“The CBO afforded me some freedom to explore,” says Smetters, who remembers using typewriters and old-school computers to run simulations and build computable general equilibrium models after he joined in 1995. Outdated technology aside, Smetters saw room for improvement: “When they’re writing legislation, policymakers want to know the budget’s impact on the economy before they put their reputations on the line.”
As he interacted with elected officials and learned about their pain points, Smetters realized that the technical side of U.S. economic modeling could also use some work. While the CBO and its tax counterpart, the Joint Committee on Taxation (JCT), provided specialized analysis for major programs like Social Security and Medicare, they lacked the bandwidth to offer more holistic modeling of how the economy impacted the programs simultaneously. “CBO and JCT were never set up to look at the major reform,” Smetters says.
After another role in D.C. in the early 2000s with the U.S. Treasury, where he navigated issues like the Enron scandal, Smetters longed to return to academia. Wharton seemed like the ideal place from which to launch an initiative that would put his D.C. political savvy to use while taking economic modeling to the next level.
Thus PWBM began to take shape in 2015, as a nonpartisan safe space in which members of Congress could check their math confidentially, with the backing of Ivy League-educated PhDs. With Smetters at the helm, a team of economists would expand CBO’s mission, as he tapped CBO colleagues Felix Reichling and Alexander Arnon to embark on this new adventure. His pitch, recalls Arnon, who’s now PWBM’s director of policy analysis, was succinct: “It’s half think tank, half startup.”
Director of policy analysis Alexander Arnon
As a small plastic desk fan circulates air in Arnon’s office and the bustle of Market Street in June flows up through the window, it’s easy to see that PWBM hasn’t deviated much from Smetters’s original vision. “I’m still kind of amazed that it worked and we’re still here,” Arnon says. As he remembers it, the team started with himself plus one economist, an administrative assistant, and a programmer. The Tax Cuts and Jobs Act of 2017 was the first major piece of legislation that flexed their muscle, leading to a citation in the Journal of Economic Perspectives.
PWBM has since grown to include several more economists and statisticians, with Arnon leading the weekly meeting for the “micro” side of the shop, otherwise known as the policy analysis team. On that summer day, there’s a buzz of cerebral energy as 10 experts file into the conference room for Arnon’s meeting and two more pop up online. The staff members, most of them bespectacled and freshly caffeinated, with mugs in hand, take turns giving updates. Acronyms such as CDC and ACS — the Centers for Disease Control and Prevention and the American Community Survey, respectively, from which PWBM gets some of its data — fly across the large table. The level of care that goes into handling such data is one of the reasons PWBM has repeat customers on both sides of the aisle.
Data is where Junghoon Lee’s team does the heavy lifting. Lee, PWBM’s soft-spoken director of engineering and data, leads his own biweekly meeting of some six engineers, who pull up lines of code onscreen to solve technical issues. In a recent virtual meeting, they plugged away at the Model Automation Pipeline, which will make it easier for PWBM’s potential external collaborators, including CBO, to run their own simulations with minimal handholding. The goal is to improve the integrity of economic modeling across the board and open up the model to others.
Director of engineering and data Junghoon Lee
“It is still quite manual,” says Lee, who joined PWBM a year and a half ago and has more than a decade of experience in software engineer and data scientist roles, spanning the pharmaceutical and energy industries, plus a PhD in electrical engineering and computer science from M.I.T. “When we automate it properly, we can improve consistency in a meaningful way.”
The pipeline includes PWBM’s version of the overlapping generations model (OLG), an economic framework that accounts for how individuals will affect the economy differently based on their stage of life. “Every year, a new group of people joins the economy, and eventually, they die,” says Lee, describing some of the variables that go into the OLG. “Companies, households, government — they all interact dynamically within the model.”
PWBM’s OLG is unique in how it factors in how household and business decisions are affected by policy changes. During Lee’s meeting, as an analyst meticulously maneuvers a prototype code, this systematic science begins to resemble a delicate art. The Model Automation Pipeline folds PWBM modules such as Social Security and the Household Tax Simulator, each with its own dependencies, into the OLG. Arnon likens the challenge of aggregating numerous individual decisions and figuring out how they interact to dominoes — “Except imagine that when a domino knocks down,” he says, “it might pick itself back up and then go the other way and knock another domino.”
There are two main deliverables that come out of PWBM. One is the classified analysis it produces for legislators, which Arnon calls its “bread and butter.” The other is the summary briefs its staff puts together — on issues ranging from tariffs to housing — and posts on its website for anyone to read.
“As part of developing our policy models, we can end up with a lot of interesting data, interesting facts about the world,” says Arnon, whose team produced one of PWBM’s most extensive summary briefs to date last December — an ambitious package of 13 major tax and spending reforms. The illustrative reforms, which include raising the Medicare age from 65 to 67 and introducing mandatory Health Savings Accounts, would raise GDP by 21 percent and reduce health insurance premiums by 27 percent, among other economic benefits. PWBM is careful not to take policy positions, and there’s no shortage of legislation in Washington to keep them busy. “Big tax reform efforts in Congress give us an opportunity to show how we can be useful,” Arnon says.
Transparency is a priority for PWBM. In certain reports, such as its 2021 analysis on the Build Back Better Act, it allows visitors to download their estimates (which can sometimes take economists up to six months to calculate) straight to Excel. One 2019 blog post about how oil prices affect business development has data stretching back to 1987.
Managing director Susan Guthrie
Balancing the open, public-service element of PWBM’s work with the ironclad trust it has built with politicians has taken time and effort. Before the pandemic, PWBM would host events on Capitol Hill to encourage interaction between policymakers and staff. Now that its prominence has grown, lawmakers seek out the PWBM. They know the drill: PWBM doesn’t disclose who it’s working with and is strictly nonpartisan. After its economists complete their analysis, legislators are given a standard template that includes the economic, budgetary, and distributional impacts of the proposed piece of legislation. “But the moment that they reference us,” says managing director Susan Guthrie, “that’s when it stops being confidential, and we will post the estimate on our website. That’s an enforcement mechanism, so that they can’t cherry-pick the parts of the analysis that they like and then just attach the Budget Model’s name to it.”
The agency that was the blueprint for the PWBM, the CBO, is now cited as a peer by journalists and the CBO itself, and word is spreading through the grapevine among the D.C. elite in both major parties. Increased awareness among policymakers and the media are the metrics by which PWBM measures its success. Smetters says the clickbait nature of today’s headlines has allowed PWBM to let its work speak for itself. “There’s a race to the bottom of who can get the press first,” Smetters says. “It’s a classic market-for-lemons problem.” PWBM’s solution is to inform legislators and voters alike.
With so many other nonpartisan watchdogs available, PWBM’s secret weapons are its in-house modeling and an unwavering commitment to accuracy, even when politicians don’t want to hear the bad news. While Republicans claimed their recent budget reconciliation bill would take U.S. economic growth to 3 percent, Smetters countered that it could maybe go up to 1.85 percent (and in a recent interview with Politico, he called some of the short-term growth effects from the bill “a bit of a mirage”). Democrats, on the other hand, seemed unhappy that PWBM didn’t show that the bill would immediately crash the economy.
“We’re loved and feared and hated by both parties,” says Smetters, leaning back in his swivel chair and confessing that he only got three hours of sleep the night before my visit because of the reconciliation-bill whirlwind. “But at the same time, they both still reach out to us. They both want to know what we have to say. The fact that we’re viewed as the honest broker — I think it’s great.”
Calling politicians’ bluff on salient topics like student-loan relief will remain one of the ongoing challenges of PWBM’s service to the American public. (Examples: PWBM said the now-defunct SAVE plan would have cost $475 billion — more than four times what was originally projected; the Inflation Reduction Act didn’t reduce inflation, Smetters points out, but also didn’t cause a ton more of it; so-called Universal Basic Income already exists, he notes: “It’s called the Earned Income Tax Credit”; and immigration “is definitely good for the U.S. economy in so many different ways, even for native-born workers,” he says.) But PWBM is already setting its sights on what’s ahead.
Expanding its accessibility and getting more people around the globe experimenting in the PWBM “sandbox” is the goal. One way to make that happen is the Model Automation Pipeline that Lee’s team is creating. Smetters estimates it will be possible for external users to use PWBM’s code base as soon as within a year. “They can’t just download it on their local machine. It’s just too big,” he explains. “But what we can do is give them exposure to a front end, and then we can basically run on our platform.” The PWBM has been contacted by foreign governments in the past, and Smetters says he would love to be able to accommodate them.
Arnon believes that smaller countries especially could benefit from PWBM’s modeling, estimating that the CBO is 10 times the size of the budget group an emerging market might have. “Most countries won’t have the resources to develop that kind of capacity, and what we can potentially provide is an infrastructure,” he says. “I’m very excited about that possibility.”
As Smetters looks ahead to the future, his vision captures the ethos of the PWBM office: realistic and cautiously optimistic. It’s clear he’s in the right place and has the experience required to provide the insights that politicians — and the public — need. “I really value the opportunity that Wharton has provided,” he says. And PWBM, propelled by the School’s community, is ready to use economic modeling to make a lasting impact on the world.
