This Working Paper presents the methodology and results of a stochastic simulation evaluation of the risks related to the medium-term world economic outlook produced using the Federal Planning Bureau’s NIME model in January 2006. The results include confidence intervals around a baseline projection for the euro area, the United States and Japan, as well as probability estimates for specific occurrences.
The baseline projections of the NIME Economic Outlooks are the outcomes of deterministic simulations of the NIME model. In a deterministic simulation, the values of the error terms of the model’s behavioural equations are assumed to be zero, while the values of the exogenous variables are set to predetermined levels. The stochastic simulation of the NIME model is based on repeated random drawings for the values of the error terms of the equations as well as for the exogenous variables of the model; for both of these types of variable the distributions are based on the historical variances of the variables’ error terms. The resulting set of stochastic runs can be presented in terms of confidence intervals around a baseline, as well as in terms of probabilities of specific occurrences.
In this Working Paper we construct confidence intervals on the basis of one thousand random drawings. The baseline that is used is the medium-term outlook presented in the NIME Economic Outlook of January 2006. The results indicate for instance that, on the basis of information available at the end of 2005, there is a 95% probability that GDP growth in the euro area will come out between 1.5% and 3% in 2006. Similarly, there is a 95% probability that real GDP growth will come out between 0.3% and 2.8% in 2012. For the United States, the 95% confidence interval places real GDP growth between 2.5 and 4.2% in 2006 and between 1.3% and 3.3% in 2012. For Japan, there is a 95% probability that real GDP growth will come out between 0.8% and 2.8% in 2006 and between 0.1% and 2.4% in 2012.
Stochastic simulation also allows us to estimate the probabilities of specific occurrences. For instance, the simulation results indicate that there is a limited 1% probability that, under a constant policy assumption, GDP growth in the euro area will be lower in 2006 than in 2005. Furthermore, the probability that GDP growth in the euro area will come out above 3% for every year of the 2010-2012 period, as aimed for under the European Union’s Lisbon Agenda, is also negligible. There is only a 13% probability that euro area inflation will effectively stay below the ECB’s 2% upper limit for every year of the 2006-2012 period. The probability that the euro area’s nominal short-term interest rate will come out at least 25 basis points higher in 2006 than its 2.2% average of 2005, is estimated at 24%. The probability that monetary policy will tighten more strongly, raising the short-term interest rate by an average of at least 50 basis points in 2006, is estimated at only 2% (these probability estimates are based on the NIME Economic Outlook of January 2006, which was itself largely based on data available in the autumn of 2005). This compares with a 84% probability that the US short-term interest rate will come out at least 50 basis points higher in 2006 than in 2005. Finally, although the baseline projection shows that the rise in Japan’s deflator of private consumption will come out at 0.3% in 2006, stochastic results indicate that there is still a 32% probability of deflation in 2006. However, the probability of deflation occurring consecutively in 2006 and in 2007 is only 5%.