8 Ian Babetskii
4. Data
Data issues in the new member states require extra attention, in particular if the objective is cross-
country comparison. Despite the apparent simplicity of the data needed, which are basic
macroeconomic indicators, no single source provides either sufficient coverage or a sufficient
length of quarterly time series. Therefore, a novel contribution of this paper is the construction of
a data-set of quarterly macroeconomic indicators for the eight NMS, namely the Czech Republic
(CZ), Estonia (ES), Hungary (HU), Latvia (LA), Lithuania (LI), Poland (PL), Slovakia (SK), and
Slovenia (SL). To have a benchmark, we also collect data for Austria (AT), Greece (GR), and
Portugal (PT). The following sources are used (in order of priority):
- Eurostat New Cronos Database;
- IMF International Financial Statistics;
- OECD Statistical Compendium and Analytical Database;
- National Statistics.
Priority was given to Eurostat data, because they have the broadest coverage of the new member
states. The data from the other sources were checked for consistency and selected so as to provide
maximum compatibility with Eurostat. There is a question as to which form of data, original or
deseasonalized, should be used in the estimations. We opted for seasonally adjusted data, since
some of the series were available only in deseasonalized form5. At the final stage of the
construction of our data-set we removed, where necessary, the seasonal component by applying
the U.S. Census Bureau’s X12 procedure, a method commonly used for seasonal adjustment6.
In short, our data-set contains nominal wages (average monthly earnings, economy wide),
consumer prices (CPI), gross domestic product (GDP) at constant prices, overall employment, and
the standardized unemployment rate (OECD measure, comparable to the labor force survey
concept). Figure 1 illustrates the development of unemployment rates in the NMS-8 and the
EMU-3 over the past decade. Period-averages are reported in Table 1. One can observe high
variation in unemployment rates across countries as well as over time. Wage and price inflation
are reported in Figure 2 and Table 2. The two variables tend to follow similar trends. Overall,
there is an apparent moderation in price and wage growth rates in the NMS, with gradual
convergence towards the EMU-3 levels being observed7. The development of real wages (CPI-
deflated) and average productivity (the ratio of real GDP to overall employment) is shown in
Figure 3. A situation where real wages grow faster than productivity suggests potential problems
on the labor markets (e.g. rising unemployment).
5 Seasonally unadjusted data are available for the Czech Republic and Poland. The use of seasonal dummies in
univariate static and time-varying estimates gives qualitatively similar results as in the case of seasonally
adjusted series. We also experimented with four lag differences, but in this case price inflation and wage
inflation become non-stationary and we cannot make an inference from the resulting estimates.
In the vector error correction specification - a method which is designed to deal with non-stationary series - the
application of seasonally unadjusted data does not affect the long-term relationships; however, the impulse
responses are more clear-cut. Overall, seasonal adjustment tends to smooth the short-term dynamics.
6 X12 is a sort of moving-average filtering procedure with time-evolving seasonal factors.
7 Notice the temporary shock to wages in Greece in the aftermath of euro adoption.