Are model-based inflation forecasts used in monetary policymaking? A case study

The process through which economic policy is conceived and decided cannot be simply described as the optimisation of a well-defined loss function subject to the constraints provided by a model of the economy. Even ignoring the forbidding difficulties of eliciting a stable and explicit loss function from real-life policymakers, the availability of a model reliably describing all the responses of a complex economy to policy interventions is hardly to be expected. Policy will therefore be made against the background of an incomplete model, lacking some policy transmission channels, subject to data revision and possibly to instability in the estimated equations, requiring continuous reassessment in the light of the available data. Drawing on the experience gained with a macroeconometric model at the Bank of Italy, in this paper we describe the uses to which such a model can be put in the policymaking process. We find empirical support for the claim that the model is used in policymaking by assessing the extent to which the monetary policy followed by the Bank of Italy in a recent episode was influenced by inflation projections that diverged from the announced targets and by trying to identify other influences that played a role. The episode considered covers the 1995-97 disinflation, when upper limits to the current- and next-year inflation rates were explicitly announced by the Governor. The empirical analysis clearly indicates a role for model forecasts of inflation in monetary policymaking. This conclusion is robust, as inflation forecasts are shown to possess explanatory power with respect to policy choices, together with a number of other factors, including lagged values of the policy instrument, lagged inflation and other, independent, inflation forecasts.