Government spending shocks are frequently identi?ed in quarterly time-series data by ruling out a contemporaneous response of government spending to other macroeconomic aggregates. We provide evidence that this assumption may not be too restrictive for U.S. annual time-series data.