6 Ian Babetskii
∆w = c1 + c2(u -u 1) + c3∆p + c4∆q +ε
(1)
t 1 2 t t-1 3 t -1 4 t t
where ∆wt = ln(wt) - ln(wt-1) , ∆pt-1 = ln( pt-1) -ln(pt-2) , ut is the natural logarithm of the
unemployment rate, and the last term ∆qt = ln(qt) - ln(qt-1) is productivity growth. Coefficient
c2 represents the responsiveness of the rate of change of wage rates to the unemployment rate and
thus characterizes wage flexibility. Negative and significant values of c2 suggest that wages are
flexible (growth in unemployment suppresses growth in wage costs). By contrast, positive or
insignificant values of wage elasticity point to the absence of wage flexibility (a phenomenon
known as hysteresis). Although nominal wages are present on the left-hand side, the coefficient
c2 measures, in fact, real wage flexibility, as price inflation is on the right-hand side as well. The
remaining part of wage growth (e.g. due to growth in import prices, etc.) is captured by the
constant term c1 . The hypothesis that real wage flexibility is different under various degrees of
exchange rate autonomy can be written as:
c 2 = c 2 + c 22 ERM
(2)
where ERM is a dummy taking one for ERM-II participants and zero otherwise. Substituting (2)
into (1) gives
δwt = c 1 + c2 (ut - ut-1 ) + c22ERM(ut - ut-1 ) + c3δPt-1 + c4δqt + εt
(3)
If wage flexibility is affected by the exchange rate regime, then the coefficient c" should be
statistically different from zero. In order to increase the power of the test, we estimate (3) for a
panel of eight NMS. Selected EMU peripheral countries such as Austria, Greece, and Portugal
serve as a benchmark. Notice that we do not include these countries in the panel estimations, in
order to keep some homogeneity. Beside an unequal degree of economic development, the NMS
and the EMU countries are characterized by different macroeconomic policies. In particular, an
autonomous (at least formally) monetary policy is a common feature of the NMS, while four
NMS are participants in the ERM-II and the other four are not. Thus, differences in wage
adjustment on the macro level could be linked with diverse exchange rate policies3. As alternative
benchmarks, we compare wage adjustment with the selected developed countries, which are
deprived of national autonomy in monetary and exchange rate policies (the EMU-3).
(ii) Under the Bayesian approach, the coefficients of regression are assumed to be random
variables. The question is which other parameters in the basic Phillips curve equation (1), except
the coefficient c2 , should be time-varying. Intuitively, if the slope of the Phillips curve can
change, the intercept should be allowed to be time-varying as well.
∆wt =c1t + c2t(ut - ut-1) + c3∆pt-1 +c4∆qt +εt (4)
3 Labor market policies and institutions such as unemployment benefits, employment protection legislation,
union coverage, the level of bargaining (sectoral versus nation-wide) etc. may affect the degree of wage
flexibility as well. Due to the main focus of this paper on macroeconomic cross-country comparison, an
assessment of these institutional effects, which typically requires microeconomic data, is beyond the scope of
this study.