Government spending, GDP and exchange rate in Zero Lower Bound: measuring causality at multiple horizons

This paper assesses the causality between government spending and gross domestic product in the United States at multiple horizons. We compare the Granger causality for normal periods (1959Q1 to 2006 Q4) with the causality for the ZLB period (2007Q1 to 2015Q4). We show that the Granger causality measures between government spending and GDP are very high and persistent in the ZLB period, but only if the exchange rate is not taken into account. When the exchange rate is taken into account, the Granger causality between government spending and GDP becomes very small and non-persistent.