Harvard Warns: AI Will Crush Tax Revenues and Worsen Social Inequality

A study by Harvard and the Brookings Institution, reported by Bloomberg, warns that AI adoption in the economy will drastically change budget revenue structures, raise unemployment, and deepen socio-economic inequality. Governments risk lower tax revenues alongside higher spending on worker support and retraining. The experts recommend starting fiscal policy reforms as early as 2026. The study models four scenarios to 2060 using the Congressional Budget Office's long-term baseline, which projects US debt rising from 101% of GDP in 2026 to 175% by 2056. In the first scenario, faster productivity growth (0.5 pp annually) slows debt growth to 136% of GDP by 2060. The second scenario, with similar productivity but benefits accruing to high-income groups, boosts tax revenues due to progressive taxation, resulting in debt at 126% of GDP. The third scenario involves a 3 million worker reduction and requires additional welfare spending. The fourth scenario, with 1 pp annual productivity growth and deeper AI penetration, displaces 6% of workers, shifts income from labor to capital, and despite higher GDP, requires substantial support measures, leaving debt at 126% of GDP. The authors call AI's fiscal impact a central economic policy question. Russian experts from Sberbank's Center for Macroeconomic Research note that capital income is taxed about 40% less than labor income, and they see similar risks for Russia, highlighting vulnerabilities in personal income tax and regional budgets.