AI Series: Can AI Fix the Deficit?

Episode Summary

Executive Summary: Brookings economist Ben Harris argues AI may boost U.S. growth roughly like the 1990s internet boom, but fiscal benefits will be partly offset by longer lifespans, lower labor-force participation, capital/labor shifts, higher safety-net use, defense spending, and higher Treasury yields. Even so, he thinks AI could modestly improve the deficit outlook, though it won’t solve the structural fiscal gap.

Main Topics: AI’s role in future productivity growth (Priority: 5/5): Harris says the most plausible AI upside is a productivity boom similar to the late 1990s internet expansion, not an immediate economic transformation. Why think tanks and policy expertise are changing (Priority: 3/5): He explains that Brookings operates in a more competitive, faster-moving environment with altered funding models and less traditional reliance by policymakers on think tanks. The U.S. fiscal outlook and debt sustainability (Priority: 5/5): The discussion centers on large and rising deficits, the widening gap between spending and revenues, and concerns that current debt dynamics are increasingly unsustainable. AI shock modeling and fiscal offsets (Priority: 5/5): Harris outlines five AI-related offsets—longevity, labor-force participation, capital share shifts, safety-net use, defense spending, and higher interest rates—that reduce the fiscal upside from AI growth. Treasury yields, term premium, and crowding out (Priority: 4/5): Both hosts emphasize that rising yields and a higher term premium may signal weakening demand for Treasuries and competition from AI-related private investment. Policy responses to restore fiscal balance (Priority: 4/5): Harris proposes a package: higher corporate revenue, capital tax reform, later retirement ages, improved tax compliance, and a target of zero primary deficits. Immigration as a growth lever (Priority: 3/5): The conversation closes with the view that rational, higher-skill immigration policy could support growth, labor supply, and innovation alongside AI.

Key Arguments: AI’s economic impact is likely to resemble the internet era more than a sudden job-destroying revolution; adoption may be fast, but economy-wide effects will take time. CBO’s baseline is conservative on AI, assuming about 2% long-run GDP growth and roughly flat long-term Treasury yields. If AI lifts growth to late-1990s levels, fiscal pressure eases substantially, potentially bringing primary deficits close to zero in a few years. That upside is partly offset by longer life expectancy, fewer workers relative to retirees, a lower labor share of income, more people relying on public programs, higher defense spending, and higher interest costs. Interest-rate effects may be the most important AI-related fiscal offset because the U.S. starts from a very high debt level, so small yield changes have large budget consequences. AI optimism should be discounted by roughly half once these real-world fiscal frictions are included. The U.S. still needs structural policy changes; AI may buy time and 1-2% of GDP in deficit reduction, but it does not eliminate the need for reform. Policy options to close the gap include raising corporate revenues, taxing capital more effectively, increasing the retirement age, and improving IRS enforcement. Rational immigration policy could improve both labor supply and long-run growth, especially if AI does not eliminate the need for non-AI workers.

Data Points: Current deficit-to-GDP: About 6% - Mark Zandi cites the current federal deficit as a share of GDP. CBO projected deficit-to-GDP: About 9% - Zandi notes the CBO’s longer-run projection under current law. Primary deficit now: About 3% of GDP - Zandi describes the current primary deficit excluding interest payments. CBO long-run primary deficit: About 2% of GDP - Used in the discussion of the 30-year outlook. Debt-to-GDP now: About 100% - Zandi characterizes publicly traded debt relative to GDP. Debt-to-GDP in 10 years (CBO baseline): About 120%-125% - Projected under current law/baseline assumptions. 10-year Treasury yield now: About 4.75% - Harris uses this as the current reference rate. CBO assumed 10-year Treasury yield: About 4.4%-4.5% - Long-term baseline assumption in the budget outlook. CBO productivity growth assumption: About 1% - Harris says CBO is not especially optimistic about AI-driven productivity gains. Cumulative nominal growth in the 1990s: 72% - Harris compares decade-by-decade growth outcomes. Cumulative nominal growth in the 2000s: 47% - Used as a benchmark in the growth comparison. Cumulative nominal growth in the 2010s: 42% - Used as a benchmark in the growth comparison. CBO projected cumulative growth, 2026-2035: 48% - Harris says this resembles the 2000s more than the 1990s. Long-run interest-rate shock in AI scenario: 35% increase in yields - Harris’s modeled AI scenario raises Treasury yields across the curve. 10-year Treasury yield in stress scenario: About 6% - Result of the modeled yield increase over the budget window. Labor force participation shock: 3 percentage-point reduction - Modeled AI-related reduction phased in over four years and then sustained. Labor share shock: New growth labor share falls from 54% to 30% - Harris says this applies to the share of new income growth, not the whole economy. Mortality assumption change: 0.7% to 2.0% annual reduction in age-specific mortality - Used to model longer longevity from AI-enabled healthcare gains. Defense spending shock: About $350 billion extra over 10 years - Initial AI arms-race assumption in the paper. Revised defense spending estimate: About $700 billion over 10 years - Harris says the estimate would likely be higher today. Older Americans age 65+ in longevity scenario: 74 million to 76 million - Illustrates the near-term increase in Social Security and Medicare pressure. Social Security Trust Fund exhaustion: 2032 - Harris cites this as a reason current-policy assumptions are not realistic. Medicare Trust Fund exhaustion: 2023 - Mentioned as part of the discussion on entitlement pressure. Corporate revenue share of GDP: From about 1.7%-1.9% down to about 1% - Harris argues there is room to raise corporate revenue by roughly 0.5% of GDP. Potential corporate revenue increase: About 0.5% of GDP - Part of Harris’s fiscal reform package. Potential non-corporate capital revenue increase: About 0.5% of GDP - From closing loopholes and modest rate increases. Potential retirement-age reform savings: About 0.5% of GDP - From raising the normal retirement age as longevity rises.

Pivotal Quotes: "I think we're going to look like the 1990s rather than the two decades that followed." — Ben Harris: Harris gives his bottom-line forecast for AI-driven growth versus the CBO baseline. "The tools we need, you need a bazooka instead of a handgun." — Ben Harris: He explains that the growing debt stock makes fiscal repair harder over time. "I think our economy a chance to grow out of this fiscal hole." — Ben Harris: His case for targeting zero primary deficits and letting growth help restore fiscal health.

Implications: AI may improve growth and revenue, but it is unlikely to erase the U.S. fiscal gap. Policymakers still need spending, tax, and immigration reforms, while markets may increasingly price higher rates and greater Treasury supply risk.