The Seasonal Cycle and the Business Cycle

This 1989 Journal of Political Economy paper by Robert Barsky and Jeffrey Miron argues that the seasonal fluctuations ordinarily filtered out of macroeconomic data before business-cycle analysis begins are themselves worth studying, because the “seasonal cycle” displays essentially the same qualitative features as the conventional business cycle – in some cases even more strongly. Using quarterly U.S. data from 1948-1985 and a model of deterministic seasonal dummies plus stochastic seasonality, the authors show that seasonal fluctuations account for more than 85% of the variation in the growth rate of real GNP (output is on average 8.08% higher in the fourth-quarter boom than in the first-quarter “recession”) while seasonal variation in prices and nominal interest rates is small or statistically indistinguishable from zero. At both seasonal and business-cycle frequencies the authors find the same stylized facts: output comoves across broadly defined sectors, production tracks sales closely with little inventory smoothing, labor productivity is procyclical, nominal money and real output are highly correlated, and quantities vary far more than prices. Because the seasonal cycle is driven by identifiable, anticipated causes – Christmas demand and the weather – rather than by unobserved “shocks,” the authors argue it offers identifying restrictions unavailable at business-cycle frequencies, and that these restrictions favor an endogenous-money, elastic-aggregate-supply, labor-hoarding account of the comovements over an exogenous-monetary-shock or pure-technology-shock account. They close by proposing, following Lucas (1977), that the close resemblance between the two cycles may point toward a single, unified propagation mechanism common to anticipated and unanticipated disturbances alike.

Robert B. Barsky | Macro Paper Warehouse

The Seasonal Cycle and the Business Cycle