Investigating uncertainty in macroeconomic forecasts by stochastic simulation

We investigate four sources of uncertainty with CPB’s macroeconomic model SAFFIER: provisional data, exogenous variables, model parameters and residuals of behavioural equations. Uncertainty is an inherent attribute of any forecast. We apply a Monte Carlo simulation technique to calculate standard errors for the short-term and medium-term horizon for GDP and eight other macroeconomic variables. The results demonstrate that the main contribution to the total variance of a medium-term forecast emanates from the uncertainty in the exogenous variables. For the short-term forecast both exogenous variables and provisional data are most relevant.