The reduced form ICM inflation equation
We derive a reduced form inflation equation for the ICM much in the same vein as for the AWM. The information set for this model is given by all variables included in the estimation of the price-unit labour cost system in Jansen (2004). The information set differs from that of the AWM on the following points: lags of changes in unit labour costs, Aulct, are used instead of lags of changes in trend unit labour costs; the changes in the wage share, Awst, the world commodity price index, Aptaw, and the GDP deflator at factor prices, Aqt, are not included; and the equilibrium-correction terms are those of the ICM,
Figure 8.6. Recursive estimates for the coefficients of the (reduced form)
AWM inflation equation
ecmpjCM and ecmulcjCM, which are derived from the estimated steady-state equations (cf. (8.10) and (8.11)).
ulc = p — 0.11u (0.02)
p = 0.91ulc + 0.09pi + t3.
After imposing valid restrictions on the general model, the final reduced form ICM inflation equation becomes:
Apt = 0.014 + 0.41Apt _i + 0.16Apt _2 + 0.03Apit _i (0.006) (0.10) (0.08) (0.01)
+ 0.06 gapt_i + 0.14Agap (0.02) (0.04)
— 0.078 ecmpt<CM — 0.031 ecmulclSY + dummies (0.016) (0.007)
a = 0.00205 (8.19)
Far(i_S)(5, 96) = 0.62[0.68] FArch(i_4)(4, 93) = 0.18[0.95] X2normaiity(2) = 0.16[0.92] Fhetx^ (20, 80) = 0.64[0.87]
Freset(1, 100) = 2.98[0.09]
We observe that the reduced form inflation equation of the ICM is variance encompassed by the corresponding AWM equation. Again, all restriction
imposed on the general model to obtain (8.19) are accepted by the data, both sequentially and when tested together. The reduced form inflation equation picks up the combined effects from the price and the unit labour cost structural equations, the latter is seen through the significant effects of Apt-1, gapt_ 1, and the equilibrium-correction term ecmulct^f in (8.19). Both equilibrium - correction terms are highly significant. If we deduct the respective means of the equilibrium-correction terms on the right-hand side, the constant term reduces to 0.6%, which is significantly different from zero with a t-value of 4.68. Figure 8.7 contains recursive estimates of the coefficients in (8.19). We note that the speed of adjustment towards the steady-state for the two equilibrium-correction terms is more stable than in the case of AWM.