THE ECONOMETRICS OF MACROECONOMIC MODELLING
The wage curve NAIRU
Without making further assumptions, and for a given rate of unemployment, there is no reason why wbq t in (5.5) should be equal to wf t in (5.6). However, there are really two additional doctrines of the Layard-Nickell model. First, that no equilibrium with a constant rate of inflation is possible without the condition w^ t = wf t. Second, the adjustment of the rate of unemployment is the singular equilibrating mechanism that brings about the necessary equalisation of the competing claims.
The heuristic explanation usually given is that excessive real wage claims on the part of the workers, that is, wq t > wf t, result in increasing inflation (e. g. A2pt > 0), while wbq t < w* t goes together with falling inflation (A2pt < 0). The only way of maintaining a steady state with constant inflation is by securing that the condition wbq t = wf t holds, and the function of unemployment is to reconcile the claims, see Layard et al. (1994: ch. 3).
Equations (5.5), (5.6), and wq = wf can be solved for the equilibrium real wage (wq), and for the rate of unemployment that reconciles the real wage claims of the two sides of the bargain, the wage curve NAIRU, denoted uw:
Thus, point (i) in Figure 5.1 is an example of wq = wq = wf and ut = uw, albeit for the case of normal cost pricing, that is, d = 0. Likewise, the analysis of a supply-side shock in the figure is easily confirmed by taking the derivative of uw with respect to mf.
In the case of & = 0, the expression for the wage curve NAIRU simplifies to
meaning that the equilibrium rate of unemployment depends only on such factors that affect wage - and price-settings, that is, supply-side factors. This is the same type of result that we have seen for the Phillips curve under the condition of dynamic homogeneity, see Section 4.2.
The definitional equation for the log of the CPI, pt is
Pt = Фчг + (1- 4>)pit, (5.9)
where pi denotes the log of the price of imports in domestic currency, and we abstract from the indirect tax rate. Using (5.9), the wedge pq in equation (5.7) can be expressed as
pq, t = (1 - Ф)piq, t,
where piq = pi — q, denotes the real exchange rate. Thus it is seen that, for the case of ш > 0, the model can alternatively be used to determine a real exchange rate that equates the two real wage claims for a given level of unemployment; see Carlin and Soskice (1990: ch. 11.2), Layard et al. (1991: ch. 8.5), and Wright (1992). In other words, with ш > 0, the wage curve natural uw is more of an intermediate equilibrium which is not completely supply-side determined, but depends on demand-side factors through the real exchange rate. To obtain the long-run equilibrium, an extra constraint of balanced current account is needed.[31]
Earlier in this section we have seen that theory gives limited guidance to whether the real-wage wedge affects the bargained wage or not. The empirical evidence is also inconclusive; see, for example, the survey by Bean (1994). However, when it comes to short-run effects of the wedge, or to components of the wedge such as consumer price growth, there is little room for doubt: dynamic wedge variables have to be taken into account. In Chapter 6 we present a model that includes these dynamic effects in full.
At this stage, it is nevertheless worthwhile to foreshadow one result, namely that the ‘no wedge’ condition, ш = 0, is not sufficient to ensure that uw in equation (5.7) corresponds to an asymptotically stable stationary solution of a dynamic model of wage - and price-setting. Other and additional parameter restrictions are required. This suggests that something quite important is lost by the ICM’s focus on the static price and wage relationships, and in Chapter 6 we therefore graft these long-run relationships into a dynamic theory framework. As a first step in that direction, we next investigate the econometric specification of the wage curve model, building on the idea that the theoretical wage - and price-setting schedules may correspond to cointegrating relationships between observable variables.