Motivated by distributional concerns raised by recent breakthroughs in AI and robotics, we ask how workers would prefer to manage an episode of automation in a task-based model, which distinguishes between automation and traditional technical progress. We show that under majority voting with the option to implement a “partial” UBI (as transfers to workers) it is optimal to tax capital at a higher rate than labor in the long run to fund the partial UBI. We show that, unlike traditional technical progress, automation always lowers the labor share in the long run, justifying distributional concerns. A quantitative analysis of an episode of automation for the US economy shows that it is optimal from the workers’ perspective to lower capital taxes and transfers over the transition. Nevertheless, this policy increases worker welfare by only 0.7% in consumption-equivalent terms, compared with a 21.6% welfare gain to entrepreneurs, because the welfare gains to workers from lower capital taxes are second-order, while the gains to entrepreneurs are first-order.

This paper establishes some stylized facts of the long-run relationship between growth and labor shares using historical data for the USA (1898–2010), the United Kingdom (1856–2010), and France (1896–2010). Performing individual country time–frequency analysis, we demonstrate the existence of long-term cycles in labor share of 30–50 years explaining a major part of the variance in the data. Further, the impact of labor share on growth changes sign with the frequency considered from negative at high frequencies to positive at low frequencies. Finally, the positive coefficient associated with the labor share at low frequencies increases over time.

US labor share has been falling since the 1970s. I show that it has not fallen as much once items that do not add to capital, depreciation and production taxes, are netted out. Recent net labor share is within its historical range, whereas gross share is at its lowest level. This effect holds for other high-income economies. The overall picture is no longer one of unprecedented, globally declining labor share. Using gross share as a proxy for net share can give misleading results. US gross share and inequality are correlated, whereas net share, the correct measure, is not.