Gunnar Bårdsen and Ragnar Nymoen 893as a necessity, since unverified changes in policy response and forecasting may
critically damage beliefs in the relevance of model analysis.
In this section we discuss some approaches to how the model of the supply side
presented in section 17.2.6 can be tested against a new and important development
represented by the New Keynesian Phillips curve.
As shown by Bårdsenet al.(2005, Chs. 4–6), three important and much used
models of inflation and unemployment are consistent with the view that wages
and prices are equilibrium correctingI( 1 )variables, while the rate of unemploy-
ment isI( 0 ). They are the incomplete competition model in equilibrium correction
form, the standard open economy wage Phillips curve, and the wage Phillips curve
with homogeneity restrictions; in other words, ICM, PCM and PCMr of section
17.4.3.Quaequilibrium correction models, these theories can readily be identified
as special cases of a VAR.
Cointegration is thus a common feature of the three models. They can be identi-
fied by their different theories about the main adjustment mechanisms at work. In
the case of the ICM, wages equilibrium correct with respect to deviations from the
wage level predicted by the theoretical bargaining model. In the case of the PCM,
by definition, wages do not equilibrium correct with respect to lagged wages, so
in this model equilibrium correction has to beindirectand via the reaction of the
rate of unemployment (see Bårdsen and Nymoen, 2008). In section 17.4.3 these
properties were shown to be relevant for the assessment of different supply-side
models for policy analysis.
We will show below that the same insight also applies to the New Keynesian
Phillips curve, meaning that its equilibrium correction implications can be tested
against the ICM or (any version) of the conventional Phillips curves. However, we
first give a brief summary of the current empirical status of the New Keynesian
Phillips curve.
17.4.4.1 The New Keynesian Phillips curve
The New Keynesian Phillips curve, hereafter NKPC, has become regarded by many
as the new standard model of the supply side in the macro-models used for mone-
tary policy analysis. This position is due to its theoretical underpinnings, laid out
in Clarida, Gali and Gertler (1999), and to the supportive empirical results in the
studies of Gali and Gertler (1999) (henceforth GG) on US data, and Gali, Gertler
and López-Salido (henceforth GGL) (2001) on euro-area data.
The hybrid NKPC is given as:
πt=af
≥ 0Et[πt+ 1 ]+ab
≥ 0πt− 1 +b
≥ 0st, (17.71)whereπtis the rate of inflation,Et
[
πt+ 1]
is expected inflation one period ahead
andstis a time series of firms’ real marginal costs. The “pure” NKPC is (17.71) with
ab=0, and represents the case where all firms (that are aggregated over) form ratio-
nal expectations. Both the pure and hybrid forms are usually presented as “exact,”
i.e., without an error term. WhenEt[πt+ 1 ]is replaced byπt+ 1 for estimation, a
moving average error term is implied. This has motivated “robust” estimation with,