Working Paper: CEPR ID: DP873
Authors: Danny Quah
Abstract: Typical analyses of trends and cycles take as given some (one) observable economic variable in whose time path a researcher wishes to find trend and cycle movements. But individual sectors and regions in aggregate economies move neither perfectly with nor independently of each other -- why is it useful to study their aggregate? Using a model for non-stationary, dynamically evolving distributions, this paper provides evidence that in the United States, regional movements that preserve their aggregate time path nevertheless have important, predictive comovements with aggregate GNP. Such predictive content cannot be understood in traditional macro models that seek the source for business cycles in aggregate productivity or monetary shocks.
Keywords: business cycle; growth; distribution dynamics; economic fluctuation; geographical region; large cross-section; stochastic kernel
JEL Codes: C33; E32; E37
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| aggregate real GNP growth (E10) | dynamics of the 0.6 quantile of the distribution of state incomes (D39) |
| maximum of the income distribution (D31) | aggregate real GNP growth (E10) |