18 Ian Babetskii
Bringing together the long-run and short-run outcomes, the following pattern emerges. In the
Czech Republic, a stable linear relationship between productivity, real wages (which grow faster
than productivity) and unemployment is possible basically due to rising unemployment. This is
rather an example of an undesired equilibrium. In Estonia, unemployment turns out to be
exogenous to the short-run adjustment, in the sense that deviations from long-run equilibrium are
closed by real wages and productivity. Moreover, unemployment is insignificant in the wage and
productivity equations, meaning that real wages and productivity move together, independently of
the unemployment situation. One cannot characterize such a labor market as flexible either, since
the variation in unemployment has no impact on real wages. The adjustment in Hungary and
Lithuania resembles the case of Estonia, except that productivity is no longer significant (i.e., the
short-run adjustment occurs via real wages, which, in turn, are not sensitive to unemployment).
5.4 Comparison with Micro-Foundations
The three alternative methods considered so far commonly suggest that real wages are inflexible
in the eight NMS. The same results apply to the three EMU members selected. While the time
series and panel methods detect several cases of wage flexibility during 1995-1999 (e.g. for the
Czech Republic, Slovakia, Lithuania, and Greece), the elasticity of wages to unemployment
becomes insignificant when considered for 2000-2004. Similarly, the time-varying estimates
indicate a deterioration of wage flexibility over time rather than an improvement. Cointegration
and error correction representations shed some light on the mechanism of labor market
adjustment: in three of the four cases where a long-run equilibrium was detected, real wages do
not react to changes in the unemployment rate; the adjustment to shocks goes rather via real
wages and productivity, the unemployment channel being insignificant. To complement the
macroeconomic analysis, let us look at the micro-based measures of wage flexibility.
Galuscak and Munich (2005) estimate the wage curve in the Czech Republic over 1993-2001.
The wage curve links real wages and unemployment at the regional level. The authors find
substantial wage adjustment for the period 1994-1996, followed by a decrease in wage flexibility
during the recession of 1997-1999. After 1999, wage flexibility did not return to the original
level, most probably due to an increase in long-term unemployment. This is what we can observe
from the time-series estimates of wage flexibility at the macro level: in the static Phillips curve
(Table 3), wage elasticity is significant and correctly signed for 1995-1999, then it becomes
insignificant for 2000-2004; according to the time-varying Phillips curve (Figure 4), wage
elasticity increased up to 1998 and has declined since 1999.
Blanchflower (2001) explores the behavior of wages in a larger set of Eastern European countries
during 1991-1997. A wage curve is found in the Czech Republic, Estonia, Hungary, Latvia,
Poland, and Slovakia. The Slovenian data do not support a significant link between wages and
unemployment. Wage data are unavailable for Lithuania. The magnitude of wage adjustment in
Eastern Europe, at the regional level, is found to be broadly similar to the estimates for other
countries (a wage elasticity of around -0.1); in Estonia and Latvia, the wage elasticity is much
higher, at about -0.5. Notice that the estimates are performed for the mid-1990s. In the author’s
opinion, “it is likely that the absolute size of these estimates will fall as more years of data
become available and full sets of region fixed dummies are included. This is generally what
happens in OECD countries.” More recent evidence seems to support this hypothesis. Iara and
Traistaru (2004) report wage elasticity for Poland in the range of -0.04 to -0.06 over 1992-1998.