the null hypothesis is the existence of a unit root, i.e. the series (at the level or at the first
difference) is non-stationary. A low p-value indicates stationarity.
Table 1
Poland: summary results for unit root tests1
Variable ADF(k)2 P-Value S/N3 Phillips-Perron(b)4 P-Value S/N
log(e) |
-3.07(1) |
0.12 |
N |
-2.38(3) |
0.38 |
N |
∆log(e) |
-6.03(0) |
0.00 |
S |
-5.59(9) |
0.00 |
S |
log(TOT) |
-1.15(11) |
0.69 |
N |
-7.14(4) |
0.00 |
S |
∆log(TOT) |
-3.10(11) |
0.00 |
S |
-16.07(3) |
0.00 |
S |
FLOW |
-2.34(1) |
0.40 |
N |
-2.27(2) |
0.44 |
N |
∆FLOW |
- 7.30(0) |
0.00 |
S |
-7.30(3) |
0.00 |
S |
log(OPEN) |
-5.92(0) |
0.00 |
S |
-5.92(0) |
0.00 |
S |
∆log(OPEN) |
-2.37(11) |
0.01 |
S |
-17.05(32) |
0.00 |
S |
log(GOV) |
-1.93(9) |
0.63 |
N |
-4.22(5) |
0.00 |
S |
∆log(GOV) |
-14.85(8) |
0.00 |
S |
-12.53(3) |
0.00 |
S |
log(NEER) |
-2.06(2) |
0.26 |
N |
-2.08(5) |
0.25 |
N |
∆log(NEER) |
-6.80(1) |
0.00 |
S |
-5.70(16) |
0.00 |
S |
log(DCRE) |
-2.02(0) |
0.58 |
N |
-1.98(2) |
0.60 |
N |
∆log(DCRE) |
-10.00(0) |
0.00 |
S |
-10.00(0) |
0.00 |
S |
1 Test equations were chosen with or without intercept and/or trend terms depending on the
P-value of the t-statistics in the test equation.
2 k is the optimal lag length indicated by the Schwartz Information criterion from a maximum
length of 12.
3 S is ‘stationary’ and N is ‘non-stationary’, as indicated by the corresponding test statistic at
5% level of significance.
4 b is the bandwidth chosen by the Newey-West bandwidth criterion.
18