Changes in fiscal policy typically entail two kinds of lags: the legislative lag--between when legislation is proposed and when it is signed into law--and the implementation lag--from when a new fiscal law is enacted and when it takes effect. These lags imply that substantial time evolves between when news arrives about fiscal changes and when the changes actually take place--time when households and firms can adjust their behavior. We identify two types of fiscal news--government spending and changes in tax policy--and map the news processes into standard DSGE models. We identify news concerning taxes through the municipal bond market. If asset markets are efficient, the yield spread between tax-exempt municipal bonds and treasuries should be a function of the news concerning changes in tax policy. We identify news concerning government spending through the Survey of Professional Forecasters. We conclude that news concerning fiscal variables is a time-varying process that can have important qualitative and quantitative effects.

Legislative and implementation lags imply that substantial time evolves between when news arrives about fiscal changes and when the changes actually take place -- time when households and firms can adjust their behavior. We identify two types of fiscal news -- government spending using the Survey of Professional Forecasters and taxes using the municipal bond market. The main contribution of the paper is a mapping from reduced-form estimates of news into a DSGE framework. We find that news about fiscal policy is a time-varying process and show that ignoring the time variation can have important consequences in a conventional macroeconomic model.

News - or foresight - about future economic fundamentals can create rational expectations equilibria with non-fundamental representations that pose substantial challenges to econometric efforts to recover the structural shocks to which economic agents react. Using tax policies as a leading example of foresight, simple theory makes transparent the economic behavior and information structures that generate non-fundamental equilibria. Econometric analyses that fail to model foresight will obtain biased estimates of output multipliers for taxes; biases are quantitatively important when two canonical theoretical models are taken as data generating processes. Both the nature of equilibria and the inferences about the effects of anticipated tax changes hinge critically on hypothesized information flows. Different methods for extracting or hypothesizing the information flows are discussed and shown to be alternative techniques for resolving a non-uniqueness problem endemic to moving average representations.