Noise Shocks and the Great Depression: A Structural Reassessment

Earlier empirical work finds that bad news unrelated to fundamentals had a prominent role in the Great Depression, suggesting that a very important element is missing in the existing analyses of the recession. We test whether noise shocks, once embedded in a general equilibrium framework, are relevant sources of variation during the Great Depression. We estimate a medium-scale DSGE model with a signal extraction on productivity on the Great Depression period. We find that noise shocks, while significant on impact, are not important sources of variation, particularly because of their low persistence: the estimated signal-to-noise ratio implies that agents quickly learn the true state of productivity. We argue that the earlier exploratory results are mainly driven by the omission of competing sources of disturbances. Our result can be read as a lower bound on the role of noise in the Great Depression—one that advances in identification through survey expectations, richer propagation via financial frictions, and departures from full-information rational expectations may revise upward.