How Might Fiscal Policy Respond to the Rise of Artificial Intelligence? (w35437)
A · Strong evidence NBER working paper by Karen Dynan, Douglas Elmendorf (former CBO director) and Louise Sheiner that applies CBO's long-term budget method to four AI scenarios with transparent assumptions; rigorous, but scenarios, not forecasts.
Summary based on the full text·Original: about 70 minutes to read
In short
NBER Working Paper w35437 (July 2026). The authors set out four illustrative long-term scenarios for AI: faster productivity growth shared proportionally, greater income inequality, job displacement, and a higher capital share. For each they estimate the effect on the US federal budget under current law and discuss fiscal responses. Their conclusion is to favour policies that work across scenarios given the uncertainty.
Key numbers
- +0.5 ppextra annual total factor productivity growth assumed in the 'rising tide' scenario
- 39 percentage pointsreduction in federal debt-to-GDP after three decades under that scenario, relative to the baseline
- 101% to 175%CBO's baseline path for US federal debt held by the public, 2026 to three decades later
- 16%rise in GDP after three decades in the rising-tide scenario
As reported by the source.
English·Web page·opens nber.org
What it reports
- Scenario 1 (rising tide): TFP growth up 0.5 percentage points a year raises GDP about 5% after a decade, 10% after two and 16% after three.
- Debt falls by 39 percentage points of GDP after three decades, offsetting about half of the projected rise from 101% to 175%.
- Most of that improvement (30 points) comes from noninterest spending falling relative to GDP, largely because retiree benefits and discretionary spending lag rising incomes; revenue adds only 2 points.
- The authors warn this gain may be partly illusory if policymakers keep historical spending-to-GDP relationships.
- With unchanged interest rates, debt would be a further 12 percent of GDP lower.
- Other scenarios cover inequality, layoffs and capital ownership, with responses on redistribution, support for displaced workers, and taxation and ownership of capital.
How they got it
Applies percentage changes from CBO's alternative-scenario analysis to CBO's extended baseline, with their own treatment of discretionary spending, plus economic reasoning on national saving and policy design.
Limits
- Scenarios are illustrative, with no probabilities attached.
- US federal budget only; no international, regulatory or state-level analysis.
- Relies on CBO's modelling conventions, which the authors partly question.
Why it matters for companies
Firms should expect fiscal policy debates on taxing capital, supporting displaced workers and redistributing AI gains. The paper shows that even the optimistic scenario leaves governments with choices that could change corporate tax and labour costs, so long-range plans built on AI productivity should include policy risk.
Summary based on the full text. Read the original for the full argument and data.Read the original on nber.org
- Tags
- Macro