AI Series: Can AI Fix the Deficit?

PodcastInside EconomicsBen HarrisSep 15

The discussion weighs whether AI-driven productivity and growth could ease the U.S. fiscal outlook against rising deficits, interest costs, and sensitivity to Treasury yields. The guest’s scenarios suggest that stronger growth could narrow primary deficits, but the estimates remain uncertain and exclude recessions, AI-related market losses, and policy changes. Speakers also discuss immigration, labor-force participation, longevity, and possible revenue measures as factors affecting growth and entitlement costs.

Artificial Intelligence Fiscal Policy Government Debt Economic Growth Interest Rates Demographics

Preview · 22 more points in the app

Read the Full Summary

Key Takeaways

The long-term fiscal outlook combines a structural gap of 2% of GDP before interest with interest spending projected at another 7%, while the deficit-to-GDP ratio is projected to rise from 6% to 9% over the next decade. The speaker presents the U.S. debt-to-wealth ratio of about 22% as more favorable than Japan’s, but says that comparison does not resolve the fiscal concern.

Interest rates are a central fiscal risk: the guest says that with debt at about 100% of GDP, even small rate changes matter, and models a 35% rise in Treasury yields across the curve, taking the 10-year yield from about 4.5% to roughly 6%. The guest also points to weakening Treasury demand and a rising term premium, warning that a 7% 10-year yield would be catastrophic; the reasons for the term-premium increase remain uncertain.

The CBO assumes about 1% average long-term productivity growth, while the speaker sees a 1990s-like outcome as increasingly plausible and cites productivity growth of 3% then versus 2% today. In the guest’s illustrative scenario, 3% productivity growth plus labor-force growth and inflation implies roughly 5.5% nominal growth; if nominal growth matches the late 1990s and other factors hold, primary deficits could approach zero in six or seven years, though the guest says a recession or AI-related market bust could reduce the improvement.

The guest estimates AI-related defense spending could add about $700 billion over ten years, while emphasizing that this is a “wild guess.” Separately, the speaker cites a bank note saying additional AI investment is crowding out public-sector investment and could change financing dynamics, with AI investors potentially paying what is required and Treasuries receiving the residual.

Yale Budget Lab research suggests labor-market churn since ChatGPT arrived has been slightly faster than during the PC and Internet expansions, but broadly similar. The speaker argues that the main risk would be labor-replacing productivity arriving faster than workers can adapt, though they do not expect that; Amazon’s roughly 25-year evolution is cited as an example of technology adoption unfolding gradually and giving workers time to adjust.

GEO-POLITICS

Defense Spending — The guest estimates that competing in the AI arms race could add about $700 billion in AI-related defense spending over the next ten years, while stressing that the figure is a “wild guess.”

“So when you go ahead and you speak with defense experts, it really feels like we're on the cusp of a lot of extra spending…”—

TECHNOLOGY

Labor Market Churn — Yale Budget Lab research suggests labor-market churn since ChatGPT's introduction has been slightly faster than during the personal computer's expansion in the 1980s and the Internet's expansion in the late 1990s, but broadly similar. The speaker says the main risk is labor-replacing productivity arriving faster than workers can adapt, though they do not expect that to happen.

“You can look at work, for example, out of the Yale budget lab where they look at labor market churn, and it looks like it's happening slightly faster than that…”—

ECONOMY

Long-Term Fiscal Gap — CBO projections imply a structural gap of 2% of GDP between non-interest spending and revenues by the mid-2050s, with interest spending adding another 7% of GDP. The guest describes the resulting 9-percentage-point gap as unsustainable.

“I think that CBO's projections for the next three decades, we get out to the mid twenty fifties, they're seeing a 2% gap between non interest spending and revenues, but…”—

Debt and Wealth — The speaker presents debt-to-wealth as a more favorable measure of the U.S. fiscal position than debt-to-GDP, citing a U.S. ratio of about 22%, compared with Japan in the mid-40s and Norway around 5%. This comparison does not resolve the separate concern about the fiscal outlook.

“And if you think about sort of national wealth, almost like a household, you know, do we have the resources to pay for the commitments we've made?…”—

Deficit Trajectory — The speakers say the deficit-to-GDP ratio is currently 6%, and CBO projections put it at 9% over the next decade. The primary deficit is 3% and may ease to 2%; they describe it as unusually high for a full-employment economy. Compared with about ten years ago, both the primary deficit and interest spending have risen by roughly 1 percentage point of GDP, leaving less room to address the fiscal problem.

“I look at the deficit to GDP, and right now that six percent, you're saying if you look at the CBO projections out over the ten year horizon, we get…”—

MARKETS

Treasury Risk — The guest says Treasury yields had risen steadily over roughly the prior half-year and interprets data from more than 200 annual auctions as showing weakened investor demand. The guest also describes a rising term premium, meaning investors are demanding higher returns, and warns that a 10-year yield of 7% would be catastrophic compared with about 4.75% at the time. The reasons for the term-premium increase are uncertain and could include weaker foreign demand, concern about the Fed, or competition from other investments.

“I am quite worried about our fiscal outlook…”—

FORECASTS

Growth and Term Premium — The guest estimates that if nominal growth matches the late 1990s and other factors do not change, primary deficits could fall to around zero in six or seven years. The guest says investors might take solace in that trajectory and the 10-year term premium could decline.

“We looked at Goldman projections. We looked at McKinsey projections on Duran Osamoglu's projections…”—

AI And Federal Deficits

Mark Sandy introduces Brookings economist Ben Harris for an AI-series discussion of how artificial-intelligence-driven growth could affect the federal fiscal outlook.

Think Tanks And AI

Harris describes changing pressures on think tanks and explains that he uses AI as a research and writing aid, while emphasizing that human judgment and writing skills remain essential.

U.S. Fiscal Risks

Harris sees a large structural budget gap and rising Treasury yields as reasons for concern, even though the United States has substantial national wealth and potential tax capacity.

Treasury Yields And AI

Harris says long-term fiscal projections depend heavily on Treasury demand and interest rates, while AI investment may compete for capital and CBO projections assume little productivity acceleration.

AI Productivity Timeline

Harris expects AI’s economic effects to unfold more gradually than some forecasts suggest, comparing the transition with earlier technologies and the long evolution of Amazon.

Growth And Primary Deficits

Harris estimates that late-1990s-style growth could bring primary deficits close to zero, but cautions that AI-specific effects could offset about half of the fiscal improvement.

AI’s Fiscal Headwinds

Harris models five countervailing forces: longer lifespans, lower labor-force participation, income shifting toward capital, increased benefit claims, and higher defense spending.

Interest Costs Temper Growth

Higher Treasury yields are the most consequential modeled fiscal headwind, and Harris says the combined assumptions reduce the projected deficit improvement from roughly four points of GDP to about two.

Deficit Outlook And Uncertainty

Harris expects AI to help but not resolve the fiscal problem, noting that the model excludes recessions and possible AI-related market losses and that trust-fund policy assumptions may not hold.

Immigration And Labor Supply

Harris says immigration assumptions materially affect growth, while the hosts argue that a more rational immigration policy could expand the workforce, support innovation, and increase demand.

Tax And Spending Options

Harris identifies higher corporate and capital-tax revenue, a higher retirement age, and stronger tax compliance as ways to move toward zero primary deficits.

Immigration And Policy Wrap

The group revisits the potential for bipartisan immigration reform, discusses the limits of the fiscal model’s policy assumptions, and closes the interview.