Don't count on AI to fix America's deficit problem
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Sarah Grillo/Axios
If the U.S. experiences an AI-fueled productivity boom over the coming years, it will mean stronger growth — but, at least under current tax policies, that would have limited benefits for the U.S. fiscal outlook, per a new analysis.
Why it matters: With the U.S. public debt high and rising rapidly, and the political will to reduce deficits by traditional means — spending cuts and tax increases — nowhere to be found, it would be awfully fortunate if America could grow its way out of its fiscal predicament.
- New modeling from the Budget Lab at Yale, however, suggests that in an AI-driven productivity boom, the fiscal situation would improve, but not by the leaps and bounds one might expect.
- That's because it would likely shift much of the national income away from human workers (labor), who the U.S. taxes at relatively high rates, and toward machines and software (capital), which is taxed more favorably.
Zoom in: The top federal income tax rate on labor income is 37%. By contrast, the corporate income tax rate is 21%, the top long-term capital gains tax rate is 23.8%, and much capital ownership occurs through tax-protected vehicles like retirement accounts and charitable endowments.
- So if AI allows companies to make more money but pay less to human labor to achieve it, those returns would not translate into the kinds of revenue gains for the federal government that accompanied past growth surges in which the labor share of national income was steadier.
By the numbers: In a scenario where AI creates only a slow boost to GDP growth, the Yale team's model shows little change in federal revenue in 2030.
- In a scenario of rapid AI-fueled growth, with 3.3% annual GDP growth in the coming years, and the labor share of income falling, federal revenue would rise by $216 billion in 2030.
- The Congressional Budget Office's baseline forecast pegs the U.S. budget deficit at $2.2 trillion that year, about 10 times the Yale team's estimate of the revenue gain from even an optimistic AI growth scenario.
What they're saying: "On the one hand, faster productivity growth would generate more tax revenue, all else equal," John Iselin and Ryan Nunn of the Yale team wrote. "On the other hand, our current tax system may not be structured to efficiently collect revenue from the economic activity produced by AI."
- "While we project that AI growth will raise revenue, without substantial changes to how the U.S. taxes capital income, the federal government will be leaving a lot of revenue on the table," Iselin tells Axios.
Reality check: These aren't the kinds of estimates that can be taken to the bank. The range of possibilities for how the AI boom will play out and how it will affect the fiscal picture is endless.
- How much will the labor share of income fall? How will it affect inequality among wage-earners (higher incomes are taxed at higher rates)?
- On the spending side, will the existing social safety net face big obligations to help displaced workers, or might there be such widespread job losses that Congress feels compelled to offer more expansive benefits than provided under current law?
- Tax policy isn't set in stone. In a world where AI is taking people's jobs and the U.S. faces a fiscal dilemma, Congress may well look to shift more of the tax burden toward capital.
The bottom line: The point is not to take the Yale Budget Lab numbers as gospel truth, but rather to show that the moving pieces between an AI-driven growth surge and federal government tax revenues aren't as linear and beneficial as one might hope when staring at an intractable deficit problem.
