specification is an EGARCH specification with suitable lags of the variables in first differences.
Comparison of equations 2, 3 and 4 in Table 8 indicates that the set of variables found significant
for the pre-crisis period is nearly the same as that for the entire period. The coefficient of the
error correction term in Specification 4 is much smaller than in specification 2 and 3, likely
because Russian monetary authorities had less liberal policies for the exchange rate in the pre-
crisis period.32 Specification 4 will be utilized in Section 9 for out of sample forecasting.
Table 7
Poland: error correction equations (equation (5)) for real exchange rate
|
Coefficients Estimates | ||
|
Variable |
Specification 1 |
Specification 2 EGARCH(7, 4) (1995:07 - 2001:09) |
|
EC(-1) |
-0.026933 |
-0.042708 |
|
(0.21) |
(0.00) | |
|
∆log(TOT) |
-0.025976 |
-0.018804 |
|
(0.04) |
(0.00) | |
|
∆FLOW |
-0.130739 |
-0.103030 |
|
(0.06) |
(0.00) | |
|
∆log(OPEN) |
0.035170 |
0.016684 |
|
(0.02) |
(0.00) | |
|
∆log(GOV) |
0.030413 |
0.013732 |
|
(0.18) |
(0.02) | |
|
∆log(DCRE) |
0. 170624 |
0.076929 |
|
(0.01) |
(0.00) | |
|
∆log(NEER) |
1.012465 |
0.938424 |
|
(0.00) |
(0.00) | |
|
∆log(TOT(-1)) |
-0.003272 | |
|
(0.00) | ||
|
∆FLOW(-1) |
-0.069762 |
32 The results on misalignment in section 8 below indicate that this rigid policy for the exchange rate contributed
to the currency crisis.
29