Government spending shocks may be anticipated before they materialize. This fact poses significant challenges to structural vector autoregressive (SVAR) analysis. The reason is that market participants have a larger information set than an econometrician seeking to investigate their  behavior based on current and past fiscal data, such that the econometrician would not be able to recover the structural fiscal shocks based on his information set. This paper shows that both
anticipated and unanticipated spending shocks can be identified if market participants’ expectations on future spending are observed. Thereby, the validity of SVAR analysis is reinstated since observing the expectations of market participants aligns their information set with the econometrician’s information set. The merits of the method are discussed based on Monte Carlo evidence from a standard model, and it is applied to the U.S. using expectations data on federal
government spending from the Survey of Professional Forecasters.