Table 3
Restrictions Tests on Model 1
As shown in Table 3, the Wald test indicates that we fail to reject the null hypothesis of
different coefficients at 90% of significance. Second, we tested in Model 1 the hypothesis
that the various costs add up to prices (cost homogeneity). We also fail to reject this null
hypothesis at a 61% of significance (Table 3). As a result we imposed both restrictions in
Model 1.
In order to evaluate the models’ forecasting ability, we estimated them until 1997:4 and
generated out-of-sample inflation forecast from 1998:1 up to 2001:1 (Figure 3). Then, we
estimated them again until 1998:4 and generated out-of-sample inflation forecast from
1999:1 up to 2001:1 (Figure 3). We find that both models follow actual inflation rather
well. However, both overestimate inflation at the end of 2000. The mean absolute error of
both forecasts is found in Table 2 and Figure 3.