Last summer, the Roosevelt Institute released a study on the economic impact of a universal basic income program that would give all U.S. adults an unconditional cash grant every month to cover basic expenses. Using a Keynesian model, the left-leaning think tank said such a giveaway would boost the economy in part because lower-income households would spend what they received. For example, it found that if all adults received $6,000 a year in grants, it would boost GDP by about 7% in the program’s eighth year, if financed by borrowing rather than raising taxes.

“When we did this exact same analysis, we found just the opposite effect,” said Kent Smetters, Wharton professor of business economics and public policy and faculty director of the Penn Wharton Budget Model (PWBM), at its first annual Spring Policy Forum held last week in Washington, D.C. Using data analytics and cloud computing, the PWBM offers a non-partisan, interactive, online tool to figure out how various changes in government policies might affect the economy. Based on the model, Smetters and his team found that GDP would fall by 6.1% by 2027 and by 9.3% by 2032. The proposed universal basic income plan would also raise the federal debt by more than 63.5% by 2027 and by 81.1% by 2032. With fewer taxes collected, Social Security revenue drops by 7.1% by 2027 and 10.4% by 2032.

On the other side of the political aisle, the Penn Wharton model also has bad news for the White House. After analyzing President Trump’s plan to boost federal infrastructure investment by $200 billion to provide incentives for a total new investment of $1.5 trillion, it disagreed with the results. The model said most of the grant programs in the plan “fail to provide strong incentives” for states to invest additional money in public infrastructure and the plan will have “little to no impact on GDP.” The White House called the analysis “flawed” and challenged the findings.

Smetters said the PWBM uses advanced modeling techniques to analyze public policy, regardless of political beliefs. “The core mission is really to move from ideology to a much more rational discussion” about the economic impact of public policy changes, he said. “We’re about analysis. We don’t advocate, we don’t recommend legislation. What we really bring is some deep theoretical modeling, data analytics, but also the software development process, which is very integral to what we do.”

Economic analyses play an important role in informing legislators as they go about voting on laws that affect every American. But too often, these analytical reports are generated by groups that have a particular point of view — whether it is a conservative think tank’s pro-tax cut agenda or a liberal consumer group’s support of expanding public benefits. Their analyses also could be too simplistic; Smetters said some can run easily on an Excel spreadsheet. So there is a need for robust, reliable, nonpartisan analysis, which is where the PWBM comes in.

“The core mission is really to move from ideology to a much more rational discussion” about the economic impact of public policy changes. –Kent Smetters

The model uses big data science, cloud computing and visualization tools in an advanced economic model to analyze public policy. Smetters said it is a “dynamic” model where users can run their own simulations of outcomes by manipulating different variables. Moreover, it uses an “overlapping generations model that incorporates all sorts of uncertainty and other very rich details that existing models — which are much more calibrated to unrelated previous policies — are not going to pick up,” he noted. This technique considers different age groups because decisions made by older generations can affect younger generations.

According to Smetters, the Penn Wharton Budget Model spends 80% of its time validating results and the rest on coding. “We go back in time. We compare census data against our model predictions. We first do that before we project forwards.” Once the microsimulation is validated, the model is used to look ahead. Another key feature is that the PWBM considers the country’s debt load in its analysis. For example, it shows that the Trump tax cuts will grow the economy about 0.06% to 0.12% a year but go down over time because of the impact of debt. “This is much below the 0.4% that was very commonly discussed in Washington” in order for the tax cuts to pay for themselves, he said.

Personally, Smetters said, it was his desire to objectively see “where we’re headed as a nation” that led to his involvement in the PWBM. He echoed the sentiments of former Microsoft CEO Steve Ballmer, who also spoke at the conference. Now retired, Ballmer said he was urged by his wife to engage in philanthropy. But Ballmer also wanted to know where the government’s finances stood, so he created a nonprofit, nonpartisan group called USAFacts. The group applies the rigors of Wall Street to the government’s finances — and came up with an annual report of Uncle Sam’s revenues, expenses and years of net losses. It only uses government data.

“The key is to try to develop a model, given the data that are available, that makes the best use of those data to make the best possible projections you can.”–Stephen Goss

Many Economic Models

Thomas Barthold, chief of staff of the U.S. Congress Joint Committee on Taxation, said it is important to remember that while economic models have improved over time, they are still “imperfect.” He added that “we don’t have perfect knowledge about the effects of deficits on the economy, in particular the timing or magnitude.” But the role of economic models remains important. “We’re trying to have both qualitative and quantitative guidance to provide to policy makers, our elected representatives who make the decisions that are important for our future,” said Barthold, who sat on the same discussion panel as Smetters.

Barthold noted that when President Kennedy instituted his round of tax cuts in the 1960s, they were based on “simple economic modeling.” Economic models also were used in decisions that led to the income tax surcharge in 1968 and 1969. And, “with the increasing deficits in the 1980s, you saw a lot of emphasis put on projections of what was going to happen to the federal deficit,” he said. However, economic models have limitations, too. During the twin energy shocks of the 1970s with oil shortages and rising prices, Barthold said, the models could not find the cause of stagflation — slow economic growth with high rates of inflation.

There is no perfect economic model, said Stephen Goss, chief actuary of the U.S. Social Security Administration, during the panel discussion. “The key is to try to develop a model, given the data that are available, that makes the best use of those data to make the best possible projections you can.” For Social Security, assumptions for its projections revolve around birth rates, immigration and the amount of revenue coming in. “In the near term, it’s all about the economy, workforce and employment,” he said. “Longer term, it’s all about demographics.”

Today, there are more organizations offering their own economic analysis. “You’ve seen accounting firms establish their own independent modeling,” Barthold said. “They often use this to talk to members [of Congress] about policy proposals.” Coalitions that advocate broad reforms now hire outside economic modelers while some think tanks have created their own broad models too, he noted. But as the crowd gets bigger, policy makers want to know which one to believe. “There becomes an increasing pressure for what’s behind the models,” he said. “What’s the underlying model? What are the assumptions? What are the data?”

“There becomes an increasing pressure for what’s behind the models. What’s the underlying model? What are the assumptions? What are the data?”–Thomas Barthold

Smetters acknowledges that while the Penn Wharton model strives to be objective, some could construe that the assumptions used in the analysis might not be seen as completely nonpartisan. “You have a default. How do you come up with those defaults?” he said. “It does require a subjective decision,” but it’s one based on academic literature. “We’ve had people from both the left and the right say, ‘you know, love your model, but … the default setting should be here or it should be here.’ And that’s fine.”

Smetters said he welcomes those arguments about economic assumptions because they are “rational debates” instead of ideological fights. “For us, the idea is we don’t have to take a stand because ultimately we want people debating about these things,” he said. “We provide the platform; you can actually change elasticities and look at different outcomes yourself.”