14 Ian Babetskii
Table 4b: Test of Similarity of Wage Elasticity across Country Groups
|
______________1995-1999_____________ |
___________2000-2004____________ | |||||
|
Chi-square |
Df |
Probability |
Chi-square |
Df |
Probability | |
|
NMS-8 |
31.05 |
7 |
0.0001 |
7.42 |
7 |
0.3872 |
|
CE-4 |
18.45 |
3 |
0.0004 |
3.16 |
3 |
0.3681 |
|
ERM-II |
9.72 |
3 |
0.0211 |
2.17 |
3 |
0.5389 |
Note: Wald coefficient test, seemingly unrelated regression (SUR) estimates of eq. (1). The null
hypothesis is that wage elasticity is the same for a given country group. No other restrictions are
imposed.
Next, the panel estimates (Table 4) performed for the group of eight new member states show
similar pattern of a decrease in wage flexibility (the elasticity changed from -0.136 for 1994-1999
to an insignificant 0.015 for 2000-2004). The effect of the ERM-II dummy has a positive sign,
meaning that wage flexibility was higher in the CE-4 compared to the ERM-II group for 1994-
1999; no significant differences are found for the recent period. Another way to test the effect of
the ERM-II group on wage elasticity is to assume that the coefficient on unemployment C2 is the
same across the eight NMS, without imposing any restrictions on the other parameters. To do so,
a system of eight country-specific equations (1) is jointly estimated. The results of the Wald test
for wage elasticity C2 are reported in Table 4b. For 1994-1999, the hypothesis of common wage
elasticity within the group of eight NMS, as well as within the CE-4 or ERM-II groups, is
rejected. On the other hand, for 2000-2004 the wage elasticity is not statistically different across
all the country-blocks. These results are in line with the time-series estimates. Indeed, during the
late 1990s, the wage elasticity in the CE-4 was higher than in the ERM-II group; in the recent
period, the wage elasticity turns out to be insignificant in all eight new member states.
5.2 Kalman Filter Estimates
Figure 4 illustrates the time-varying estimates of the slope of the Phillips curve. Wage flexibility
is indeed not constant, and there are cases of both positive and negative reactions of wages to
increases in unemployment. Positive values correspond to a sort of hysteresis, i.e., an absence of
wage flexibility. The more negative is C2t, the more flexible are real wages. Downward-sloping
patterns of wage flexibility are observed for the Czech Republic, Latvia, and Poland. However, in
the Czech case, wage flexibility was increasing only till 1998, and one can observe a decline in
flexibility from 1999 onwards. In Latvia, despite a declining trend, the wage elasticity is still
positive at the end of 2004, meaning that growth in unemployment is accompanied by growth in
wage costs. This indicates an absence of wage flexibility. Similarly in Poland, there are some
signs of wage flexibility during 2001-2003, but overall the wage elasticity is not significantly
different from zero.
For some CE-4 countries, we can tentatively observe a very weak link between the fixing of the
exchange rate and wage flexibility. For example, in the Czech Republic the exchange rate regime
changed from a fix to a float in the middle of 1997. This corresponds to an observed decrease in
real wage flexibility two years later. Similarly, the relaxation of the exchange rate peg in Slovakia
in the second half of 1998 is followed by a decrease in real wage flexibility from 2000 onwards.