Statistical Methods
The business cycle

Common elements in `international business cycles' may reflect economic interdependencies or common disturbances, and demand or supply factors can induce either, but the identification of their sources and means of transmission has received little dynamic analysis in an international context. In Discussion Paper No. 781, Research Fellow Fabio Canova develops a three-country general equilibrium model, which distinguishes contributions of demand and supply to output fluctuations and quantifies the roles of intermediate and final goods in their transmission. Demand shocks reflect government expenditure, while supply shocks reflect technological changes; their channels of transmission include propagation via trade in intermediate and final goods.

Canova first simulates the transmission mechanism to reproduce the propagation of actual disturbances. With Germany and Japan resembling the US to capture short-run features, both demand and supply shocks involve contemporaneous correlation of shocks, but with common elements in the long run, production interdependencies relate to technology shocks and consumption interdependencies to expenditure shocks. With heterogeneity, propagation of expenditure shocks is almost entirely due to trade in consumption goods, but the data fit does not significantly improve. Finally, with parameters varying across countries, technology changes best account for the transmission of US output shocks and expenditure shocks for that of German output shocks.

Canova suggests that allowing for different national monetary policies, explicit modelling of terms-of-trade shocks and different national labour market structures might all improve the model's fit. These results suggest that European financial integration should entail little change in cyclical patterns across countries if their business cycles are related like those of the countries modelled here; curbing trade may not stabilize domestic fluctuations and may reduce consumer welfare, while fiscal coordination will have little effect on medium- or long-run impulse responses.

In Discussion Paper No. 782, Canova uses various US macroeconomic time-series to examine the implications of alternative detrending methods for quantitative and qualitative business cycle `stylized facts'. He finds that the choice of `statistical' or `economic' approaches significantly affects the properties of the estimated cyclical components for GNP, consumption, investment, hours worked, the real wage, productivity and the capital stock. Qualitative responses of consumption, investment, hours worked and the real wage to a GNP shock exhibit two distinct patterns. In the first, a temporary output rise raises labour demand, hours and the real wage; consumption increases and investment follows. Average productivity increases more than the real wage, so profits and real returns to capital also rise. Hours thus move together with real returns, which is consistent with the real business cycle literature's emphasis on the intertemporal substitution of labour, while the approximate coincidence of productivity responses with those of GNP runs against explanations based on labour hoarding. In the second, an output shock leads to a larger rise in consumption, reduces hours worked, depletes the capital stock and leads to negative investment. At least initially, labour productivity is negatively related to and lags output, which is consistent with labour hoarding. A demand-driven expansion further raises output in the short run, which drives up hours and real wages; once this consumption boom is exhausted, agents enjoy increased leisure, hours worked fall below their long-run path, and the deterioration of the capital stock is reversed. Despite large movements in interest rates and real wages, hours worked move only by small amounts, which is consistent with recent neo-Keynesian accounts of the business cycle.

Canova concludes that these results demonstrate the need to reconsider theoretical models of the business cycle whose numerical versions quantitatively match `stylized facts'. Numerical exercises should provide alternative results obtained with different business cycle periodicities, while theoretical work should consider why the latter critically affect certain variables' behaviour.

Sources and Propagation of International Business Cycles: Common Shocks or Transmission?
Detrending and Business Cycle Facts
Fabio Canova