Palgrave Handbook of Econometrics: Applied Econometrics

Gunnar Bårdsen and Ragnar Nymoen 897

sharesthas the correct positive sign and is significant at the 10% level. By and
large, these results are in line with the typical NKPC Results 1–3 discussed above.
The instruments includeulct− 1 −pt− 1 andulct− 1 −pit− 1 , as explained above,
together with lags of productivity growth, lagged electricity price growth, the same
wage dummy as in NAM and lagged unemployment. The Sargan (1964) test of
instrument validity(χS^2 (n)) is insignificant. However, residual misspecification tests
reveal that (17.78) is not a congruent model, since there is substantial heteroskedas-
ticity, autocorrelation and non-normality. In order to obtain a more congruent
NKPC model we may use the dummy saturation technique in Autometrics (see
Doornik, 2008).

pt = 0.622

(0.102)

pt+ 1 + 0.02509

(0.013)

st+ 0.2614

(0.0533)

pt− 2

+ 0.0549

(0.00675)

pet+constant and 11 dummies

IV,T =111 (1979(3) - 2007(1))

χs^2 ( 15 )=19.442[0.1944].

(17.79)

The 11 dummies include both seasonals and break dummies, showing that the
NKPC equation is not a completely time-invariant structural model. However,
abstracting from that problem, equation (17.79) is seen to adhere even closer to
the stylized facts of the NKPC than equation (17.78).
As explained above, we want to test the hypothesis that (17.79) encompasses the
price equation of the incumbent model. To do that, we first includeulct− 1 −pt− 1
andulct− 1 −pit− 1 as explanatory variables to obtain an empirical version of the
embedding equation (17.77) above. We then do a general-to-specific search by
means of Autometrics in PcGive. The preferred model is reported as equation
(17.80):

pt= 0.003849

(0.069)

pt+ 1 + 0.08832

(0.0259)

ulct+ 0.1515

(0.0577)

pt− 2

+ 0.09883

(0.0143)

(ulct− 1 −pt− 1 )−0.01943

(0.00478)

(ulct− 1 −pit− 1 )+0.05164

(0.00518)

pet

+constant and 4 dummies

IV,T =111 (1979(3) – 2007(1))

χS^2 ( 16 )=25.434[0.0625]. (17.80)

It is seen that the estimated forward coefficientαˆf is practically zero in (17.80)

compared to 0.62 in the NKPC in (17.79). HenceH 0 c:αf=0 cannot be rejected
statistically at any meaningful level of significance.

17.4.5 Forecasting for monetary policy

A hallmark of modern and flexible inflation targeting is that the operational tar-
get variable is the forecasted rate of inflation (see Svensson, 1997). One argument