TL;DR

As ChatGPT revives Andrew Yang’s universal basic income arguments, economic analysis suggests the proposal fundamentally fails to address AI-driven job displacement. A $1,000 monthly “Freedom Dividend” would leave families below poverty, while meaningful income replacement would cost $14tn annually—45% of US GDP.

The Scale Problem

Yang’s 2020 “Freedom Dividend” proposed $1,000 monthly for every American adult. For a two-parent, two-kid family, this puts income 25% below the poverty line. A genuinely meaningful UBI—equivalent to median worker earnings of $53,000 annually—would cost over $14tn, roughly 45% of US GDP.

For context, US public social spending covering health, pensions, disability, and unemployment has never exceeded 25% of GDP since 1980. Total government tax revenue at all levels has never reached 30% of GDP since the 1960s.

The Funding Paradox

Yang suggested a value-added tax to fund UBI. While VATs can raise substantial revenue and don’t discourage work, the logic becomes circular in an AI economy: funding people’s livelihoods through a tax on what they buy, when they have no income to buy with.

If AI truly eliminates labour income—which currently generates most tax revenue—funding becomes even more challenging. Alternatives might include carbon taxes, land taxes, or taxing robot owners. But this raises fundamental questions about power: who decides how much everyone receives?

The Inequality Question

Stanford economist Erik Brynjolfsson warns that in an economy where labour’s income share reaches zero, capital owners “reap it all.” Most people would “depend precariously on the decisions of those in control of the technology,” risking being “trapped in an equilibrium where those without power have no way to improve their outcomes.”

UBI, as currently conceived, doesn’t address this concentration of power. Redistributing income may be insufficient—redistributing ownership of the AI systems themselves might be necessary.

Alternative Approaches

The analysis notes that other developed nations already work significantly fewer hours without UBI. Australians work 20% less than Americans; Danes and Finns work 24% less. These countries rely not on universal payments but on “a halfway decent social safety net.”

For the present—where the challenge is low-wage service jobs rather than zero employment—wage subsidies like an improved earned income tax credit may prove more effective and affordable than universal payments.

Looking Forward

The debate highlights that AI’s economic disruption may require solutions more fundamental than simple cash transfers. For UK businesses and policymakers, the questions of AI ownership, wealth concentration, and power distribution will likely grow more pressing as automation advances.