Table 5(A). Long run PPP coefficient estimates (CPI)
MG_______FE |
POLS | |
i)DM____________________________________________ | ||
ADF P [SE1] |
-.0377[.0037] -.0237[.0029] |
-.0127[.0018] |
ADF(Wτ ) | ||
P [SE1] |
-.0398[.0036] -.0247[.0024] |
-.0131[.0015] |
ARDL Y [SE1] |
-.01967[.0069] -.0044[.00079] |
-.0045[.00071] |
ARDL(Wτ ) | ||
Y [SE1] |
-.01989[.0087] -.0041[.00066] |
-.0041[.00059] |
ii) US$__________________________________________________________ | ||
ADF P [SE1] |
-.0256[.0017] -.0225[.0028] |
-5.96e-5[.00021] |
ADF(Wτ ) | ||
P [SE1] |
-.0274[.0019] -.0245[.0016] |
-4.86e-5[.00012] |
ARDL Y [SE1] |
-.0064[.0056] -.0025[.0014] |
-.0034[.0012] |
ARDL(Wτ ) | ||
Y [SE1] |
-.0024[.0030] -.0024[.00077] |
-.0040[.00067] |
Note: Tables 5A-5B report the estimation results for two dynamic PPP equa-
tions, ADF and ARDL. The number of augmentation lags is conservatively set at
k=6 in all equations to eliminate serial correlation. ADF(Wτ) and ARDL(Wτ)
denote the models with τ factors as additional regressors. The conventional s.e.
for the MG, POLS and FE estimators are in brackets. These are likely to be biased
downwards in all regressions.
27