ABSTRACT
Before discussing conflicts between forecasting and policy analysis, it is important to define the two activities clearly. The definition focuses on outputs, rather than the means of producing those outputs. A forecast is an unconditional prediction of what will happen to the economy. Policy analysis is more diverse, but it includes a ceteris paribus analysis of the effects of changing policy instruments, and the design and consequences of alternative policy regimes. Now of course the two activities can be linked: the government is part of the economy, and so a forecast can involve an analysis of the effects of government actions. However, the two activities need not be linked at all: a VAR model producing forecasts need not even include policy variables, and even if it does, it is very unclear how to identify policy shocks using it.
