difference) of each variable for each country is determined with suitable unit root tests.21 Then,
given the unit root tests, the long run equilibrium relationship, equation (4), is estimated. Then
the error correction equation (5) is specified for each country.
All data are from International Financial Statistics (IFS) and the Institute for Economic
Research, Halle, Germany. The individual variables and the corresponding data sources are
discussed in the appendix. Monthly data from January 1995 to December 2001 are employed
for each country. By 1995 the economies were more stable and more reliable data were
available.22 During the early transition period, prior to 1995 it is widely argued that the
currencies were significantly undervalued. Beginning the sample in 1995 avoids this period of
instability and the real currency appreciations that prevailed during the early period in most
transition economies.23 In addition the exchange rate regime in Poland was less flexible prior to
1995. In 1995 a managed float was introduced and in 1996 a very wide band (15%) was
introduced and the regime gradually became a free float.24
The results from Augmented Dicky-Fuller and Phillips-Perron unit root tests are presented in
Table 1 for Poland and Table 2 for Russia.25 The tests are conducted for the series in levels, as
well as first differences. All the variables, except FLOW, are in logarithmic terms. For both tests,
21 As mentioned earlier, each variable in the long run equilibrium relationship has to be either I(0) or I(1).
22 Further, data for several of the variables we include in the model below were not available until 1995.
23 The causes of this appreciation are debatable. Liargovas (1999) discusses several potential causes for the
appreciation. Dufrenot and Egert (2005) look at the determinants of the real exchange rate for Hungary and Poland
with data from 1992:1 to 2002:12 and the Czech Republic, Slovakia and Slovenia for 1993:1 to 2002:12. They
concluded that macro fundamentals were an important factor in the real appreciations that took place in these
countries.
24 See Orlowski (2000) for a brief description of the exchange rate regime in Poland at this time.
25 ADF tests are expected to perform satisfactorily even when the number of observations is small (Hamilton
(1994)). However, Perron (1989) argued that structural breaks in the data would invalidate the conventional unit root
tests. Therefore we use the Phillips-Perron test to confirm the ADF tests. For details see Hamilton (1994).
17