Adolfson, Malin Laséen, Stefan Lindé, Jesper Villani, Mattias
Year of Publication:
Sveriges Riksbank Working Paper Series 180
This paper uses an estimated open economy DSGE model to examine if constant interest forecasts one and two years ahead can be regarded as modest policy interventions during the period 1993Q4-2002Q4. An intervention is here defined to be modest if it does not trigger the agents to revise their expectations about the inflation targeting policy. Using univariate modesty statistics, we show that the modesty of the policy interventions depends on the assumptions about the uncertainty in the future shock realizations. In 1998Q4-2002Q4, the two year constant interest rate projections turn out immodest when assuming uncertainty only about monetary policy shocks during the conditioning period. However, allowing non-policy shocks to influence the forecasts makes the interventions more modest, at least one year ahead. Using a multivariate statistic, however, which takes the joint effects of the policy interventions into consideration, we find that the conditional policy shifts all projections beyond what is plausible in the latter part of the sample (1998Q4-2002Q4), and thereby affects the expectations formation of the agents. Consequently, the constant interest rate assumption has arguably led to conditional forecasts at the two year horizon that cannot be considered economically meaningful during this period.
Forecasting Monetary policy Open economy DSGE model Policy interventions Bayesian inference