THE ECONOMETRICS OF MACROECONOMIC MODELLING
. Summary of time varying NAIRUs in the Nordic countries
In sum, for all three countries, we obtain stable empirical wage equations over the period 1964-94 (Denmark 1968-94). Nor do we detect changes in explanatory variables in the wage-setting that can explain the rise in unemployment (as indicated by absence of an increasing trend in the AWSU indicator in Figure 6.4). The instability of the NAWRU estimate appears to be an artefact of a mis-specified underlying wage equation, and is not due to instability in the wage-setting itself. Note also that the conclusion is not specific to the NAWRU but extends to other methods of estimating a time varying NAIRU: as long as the premise of these estimations are that any significant changes in the NAIRU is due to changes in wage (or price) setting, they also have as a common implication that the conditional wage equations in Table 6.2 should be unstable. Since they are not, a class of models is seen to be inconsistent with the evidence.
The results bring us back to the main question: should empirical macroeconomic modelling be based on the natural rate doctrine? The evidence presented
in this section more than suggests that there is a negative answer to this question. Instead we might conclude that if the equilibrium level of unemployment is going to be a strong attractor of actual unemployment, without displaying incredible jumps or unreasonably strong drift, the dichotomy between structural supply-side factors and demand-side influences has to be given up. In the next section, we outline a framework that goes beyond the natural rate model.