New Study Finds Automation-Led Growth Hinges on Financing Capital Demand as Much as Technology

As automation enables machines to take on a larger share of the economy’s tasks, the sources of growth become more nuanced than simply “better technology.” In a new analysis built on a standard task-based economic model, the key question is not only how quickly automation improves what machines can do, but also how the economy finances the resulting demand for capital.

The paper argues that growth during an automation wave depends on two separate mechanisms. First is the task structure: how work is divided between labor and machines, which determines how profitable it is to add more capital as automation expands. Second is the capital-supply response: how strongly investment and saving rise when returns to owning machines increase.

Within this framework, the author finds that the availability of capital is what largely determines the size of the growth surge associated with automation. By contrast, the way tasks are organized between workers and machines is what shapes the length of that surge—and how wages behave along the way. In other words, two economies with the same automation technology can experience very different outcomes depending on how much additional capital they are able to accumulate, and on the specific set of tasks that machines can substitute for.

The quantitative results are stark. Following the same underlying technological trajectory, the model produces an annual growth rate of 11.7% when households are allowed to choose their saving behavior optimally in response to changing returns. But if the saving rate is held constant at a calibrated U.S. 2025 level—23%—growth drops to just 2.2% per year. The implication is that an economy that looks “stuck” might not be facing a lack of technological opportunities. Instead, weak growth could reflect insufficient capital accumulation: too little saving and investment to fully exploit automation’s high returns.

Notably, the model does not attribute higher aggregate saving mainly to households suddenly becoming more frugal. Rather, the saving rate increases primarily because automation shifts income toward capital—raising the capital share—mechanically generating more resources flowing into capital accumulation.

Financial markets, in the model, react quickly to the prospect of an automation-fueled boom. Long-term asset prices incorporate the stronger growth outlook: the ten-year real yield rises sharply on impact, from 5.1% to 13.6%. That jump reflects a world in which returns to capital increase substantially as machines become more productive and more widely deployed.

Distributional consequences also feature prominently. In the benchmark calibration, “hand-to-mouth” households—those with little or no wealth who rely on current income—fare poorly during the transition. They experience welfare losses equivalent to roughly two-fifths to two-thirds of their consumption. The paper also estimates that policies designed to keep these households’ consumption on its previous trend are not prohibitively expensive in the short run, with a temporary cost of no more than one-fifth of GDP.

Why it matters

If these results hold, debates about slow growth may be misdiagnosing the problem: the binding constraint could be financing and capital accumulation rather than a shortage of technological progress. The findings also highlight a potential policy tradeoff—automation-driven growth may be rapid, but the transition could impose large costs on liquidity-constrained households. What to watch next is whether saving, investment, and capital markets can expand fast enough to match automation’s potential—and how governments handle the distributional fallout.