Jesús Fernández-Villaverde, the Howard Marks Presidential Professor of Economics at the University of Pennsylvania, says automation’s growth payoff hinges on saving, while his model assigns substantial losses to households living hand to mouth.
Marginal Revolution published an excerpt of his research Oct. 20. In one simulation, annual growth reaches 11.7% when households choose saving optimally. With saving fixed at the model’s 23% U.S. benchmark for 2025, the same technology path yields 2.2% growth. Both rates are modeled outcomes, not recorded growth.
The benchmark calculation assigns households living hand to mouth a loss equivalent to between two-fifths and two-thirds of their consumption. That is a modeled loss measure, rather than an observed decline in spending. Fernández-Villaverde estimates that keeping their consumption “on trend” temporarily would cost no more than one-fifth of gross domestic product.
Automation raises capital’s share of income in the model. Fernández-Villaverde says that shift is the main reason saving increases. He argues that the resulting accumulation of capital accounts for much of the difference between the two growth outcomes, despite their shared technology path.
The excerpt also puts the model’s 10-year real yield at 13.6% when the automation shock occurs, up from 5.1%. Those are simulated yields. The U.S. Treasury recorded an observed 10-year par real yield of 2.77% on Oct. 20.
The saving benchmark requires similar care. The Bureau of Economic Analysis reported a U.S. personal saving rate of 5.4% of disposable personal income for 2025. The published excerpt does not define the denominator for the model’s 23% figure, so the two rates cannot be compared directly.
