Comment on “The Intertemporal Keynesian Cross”, by Auclert, Rognlie, and Straub (2024)

• Authors: Jaime Gimeno-Ribes
• BSE Working Paper: 1596 | September 2026

This comment revisits the empirical estimates of consumption responses to lotteries in Norway that are used in Auclert, Rognlie, and Straub (2024) to discriminate between competing business-cycle models. The original estimation procedure overweighs the responses to the largest lotteries, which, given the size variation and response heterogeneity across size, produces biased full-sample estimates. A population-weighted average that partially corrects for the bias yields a more frontloaded consumption response, implying that the lagged effects are at least 4x smaller than previously thought. Even after the partial correction, size considerations and the consumption imputation procedure limit the estimates’ relevance for business cycles.