Correlations between macroeconomic cycles in the US and UK: what can a frequency domain analysis tell us?

In this paper, the relationship between the growth of GDP and real GDP components is explored in the frequency domain using both static and dynamic techniques derived from wavelet analysis. This analysis is carried out separately for the US and the UK using quarterly data, and the results are found to be quite different for both countries. One of the major findings of the paper is that there are significant relationships between cycles of the GDP components at frequencies other than the business cycle frequencies that confirm some of the predictions of economic theory; but importantly contradict others. Secondly, that the relative power of the different cycles varies over time leading to the idea of ``volatility transfers'' between cycles. Third that the correlations and lead-lag relationships between cycles shift over time, leading to several new insights into how the aggregate demand in these two economies has actually evolved in practice. These results are of particular significance for understanding the current weak economic recovery and for why real business cycle models do not find the right relationships between GDP components.