Is Empirical Granularity High Enough to Cause Aggregate Fluctuations? The closeness to Gaussian

Author Name ARATA Yoshiyuki (Fellow, RIETI)
Creation Date/NO. April 2022 22-E-039
Research Project Heterogeneity across Agents and Sustainability of the Japanese Economy
Download / Links


Recent studies (e.g., Gabaix (2011)) argue that because of the high heterogeneity of firm size, microeconomic shocks generate aggregate fluctuations (i.e., the granular hypothesis). This paper tests whether empirical granularity is high enough to explain fluctuations in the GDP growth rate using firm-level data in G7 countries. I find that even when the central limit theorem does not hold, microeconomic shocks cancel each other out, and thus, the distribution of aggregate output induced by microeconomic shocks is very close to a Gaussian. In other words, the observed heterogeneity of firm size in all G7 countries is not high enough to prevent the averaging effect of microeconomic shocks. Furthermore, because of the closeness to a Gaussian, microeconomic shocks with export/import relations would result in no tail dependence of aggregate fluctuations. Since the empirical GDP growth rates deviate from a Gaussian in the tail region and show positive tail dependence across countries, the granular hypothesis cannot explain these tail features of the GDP growth rates.