Euro Adoption in Central and Eastern Europe
Chapter

3 Real Convergence, Economic Dynamics, and the Adoption of the Euro in the New European Union Member States

Author(s):
Susan Schadler
Published Date:
April 2005
Share
  • ShareShare
Show Summary Details
Author(s)
Christian Thimann*

Real convergence—generally understood as income catching-up—is arguably the most important challenge for the new member states and one of the great opportunities of an expanded European Union (EU). In addition, adoption of the euro is one of the main policy objectives of the new member states and an obligation of EU membership. What is the link between real convergence and adoption of the euro? Some observers have implicitly or explicitly argued that the large income gap impedes monetary integration and that real convergence to certain levels would be needed before new member states could participate in the euro area.

This paper takes issue with this view and argues that the difference in income levels itself is not directly relevant for monetary integration. What matters is the potentially significant difference in economic dynamics that is entailed in the real convergence process. Given that real convergence will go hand in hand with the adjustment of economic structures (inter alia, the size of the private sector, the distribution of relative sectoral employment, and similarity in production structures), it is likely to be accompanied by economic dynamics that differ significantly from economic dynamics in the current euro area. If this is the case, macroeconomic stabilization becomes an issue; there would need to be sufficient policy instruments and shock absorbers to ensure macroeconomic stabilization.

This paper highlights three elements in economic dynamics that are relevant: (1) new member states should experience in the real convergence process higher average real growth rates, (2) they are also likely to experience higher volatility in the economic cycle, and (3) peaks and troughs are likely to be less correlated with the current euro area than they would be if income levels and economic production structures were already highly assimilated. The key argument of the paper is that these three elements establish the link between real convergence and monetary integration. In other words, differences in income levels are neither directly relevant nor irrelevant for monetary integration; they are relevant insofar as they determine differences in economic dynamics that could affect macroeconomic stabilization in a monetary union.

This paper starts by recalling real income differences between new member states and the euro area as well as a few features of differences in economic structures and then provides a framework to study the link between these differences and economic dynamics. It shows that indeed economic dynamics have been quite different for some new member states compared with the euro area over the past few years. At the same time, other new members have economic dynamics that are already relatively close to those of some of the EU15 countries. Taken together, the findings of this paper recall the issue of macroeconomic stabilization for the specific situation of catching-up economies and underline the fact that monetary integration warrants a case-by-case approach.

The Real Convergence Gap

As is well known, gross domestic product (GDP) and income levels in the new member states are still far below those in the euro area, with GDP per capita at around 50 percent of the euro area average in purchasing power parity (PPP) terms, ranging from 40 percent in Latvia to 79 percent in Cyprus (Figure 1).1 Even if the strong economic expansion experienced over the past decade, with real GDP growing on average at 3.6 percent (weighted), compared to 2 percent in the euro area, were to be extrapolated, the new member states would reach 75 percent of the income level in the EU15 in only about 25 to 30 years.

Figure 1.Per Capita GDP in PPP Terms

(As a percent of the euro area average)

Sources: Eurostat and European Commission.

1Weighted by projected nominal GDP in 2004.

In line with the economic expansion over the past decade, economic structures in the new member states, by which is meant inter alia the size of the private sector and relative sectoral employment, have been continuously adjusted toward those in the euro area. Moreover, the new member states have advanced in terms of institutional convergence. Although having an economic structure similar to the euro area’s is far from being a sufficient condition to receive a net benefit from European Monetary Union membership, it may help in the country’s capacity to absorb economic shocks and to face similar shocks.2 Finally, EU membership is expected to further foster convergence, inter alia due to the continued implementation of the acquis communautaire and the stronger coordination through the surveillance procedures laid down in the treaty.

This view is broadly in line with the European Bank for Reconstruction and Development (EBRD) transition indicators, which confirm notable progress in the areas of privatization and liberalization of markets and prices in recent years.3 Some differences exist across countries with respect to reforms in enterprise privatization and restructuring, and even more so with regard to the financial sector. In contrast, progress in the area of markets and competition is more homogeneous across all eight former transition countries (Figure 2).

Figure 2.Institutional Reform, 2003

(Index: 100 = well-functioning market economy)

Sources: EBRD and author’s calculations.

Finally, the economic size of the three broad sectors and the distribution of employment across sectors have gradually converged toward those in the euro area, despite large differences across countries (Table 1). This evidence has to be interpreted with caution, however, since it does not take into account differences that may exist at a more disaggregated level. At the broad sectoral level, the current GDP shares of agriculture and industry are still higher in the new member states than in the EU15 countries, reflecting the ongoing restructuring process, while the services sector is somewhat smaller.

Table 1.Economic Size and Labor Distribution of Sectors, 20031
Economic Size as Percent of GDPEmployment Distribution as Percent of Total
AgricultureIndustry and

Construction
ServicesAgricultureIndustry and

Construction
Services
Cyprus4.320.275.65.123.171.8
Czech Republic3.438.458.24.539.755.8
Estonia4.428.167.56.931.262.0
Hungary3.730.266.15.831.962.3
Latvia4.322.872.913.325.960.8
Lithuania6.231.961.917.828.054.2
Malta2.828.169.11.830.168.1
Poland3.030.766.426.825.347.9
Slovakia3.932.064.05.838.455.8
Slovenia3.135.461.411.037.052.0
AC-103.431.864.715.930.453.7
Greece6.622.970.515.324.260.4
Portugal3.726.769.512.232.455.3
Spain3.228.568.45.929.464.7
EU2.026.671.43.928.267.8
Sources: European Commission and Eurostat.

Moreover, countries display large differences with respect to sector shares and employment distribution. In particular, Poland, Lithuania, Latvia, and Slovenia all have a considerably larger share of employees in the agriculture sector than the average for the euro area, while the relative economic outputs are more similar, pointing to much lower agricultural productivity in these new member states than in the euro area.4

Real Convergence and Structural Differences in Economic Dynamics

Real convergence is one of the key economic characteristics of the new member states. While differences in income levels and some differences in economic structures can, in principle, be compatible with participation in a monetary union, such differences can have important implications for the appropriateness of the single monetary policy for individual members. In particular, real convergence is likely to be reflected in differences in economic dynamics between members, which are increasingly regarded as leading to potential costs in joining a monetary union. Inappropriate macroeconomic policies would exacerbate periods of overheating or downturns, would lead to boom-bust cycles, and might bring about higher average unemployment over time through labor market hysteresis. As such, if differences are substantial and persistent, abandoning an independent monetary policy as a stabilization instrument may lead to welfare losses. This section explores these considerations for the new member states.

Over the past 13 years, economic growth in most of the new member states has developed quite differently from economic growth in the euro area, with the exception of Cyprus and Malta.5 The following analysis examines the basic properties of output dynamics since 1996 (the first half of the 1990s, by contrast, was shaped mostly by systemic transformation) and aims to assess structural differences vis-à-vis the euro area.6 Note that owing to the short time series, all findings have to be interpreted with caution and may be subject to revisions when new data become available.

Three important features stand out, all of which affect the choice of timing of euro area entry: growth rates have been persistently higher in the new member states, growth fluctuations (i.e., amplitudes of upswings and downturns) have been more pronounced in the new member states, and new member states’ business cycles have not always been closely synchronized with those of the euro area.

A convenient way of condensing these differences into a single indicator is Theil’s inequality coefficient. This coefficient measures a scaled root mean squared difference between two series. It takes values between zero (perfect fit) and unity. The statistic provides two important benefits. First, it allows the comparison of different pairs of variables at different scales, with respect to a broad concept of inequality. Second, inequality of time series can be broken down into its main statistical factors: difference in mean, difference in variability, and lack of correlation. This method for measuring inequality was introduced by Henri Theil in 1967. It has been widely used in studying the welfare economics of income distribution and has been more recently also applied to time series. In this setting, Theil’s U inequality coefficient measures the degree of differences in dynamic patterns of two time series. Specifically, the inequality coefficient between two series, y and x, which have T observations, is defined as their scaled root mean squared difference:7

where the numerator represents the mean squared difference and the denominator provides a scaling factor. As a result of the scaling factor, the value of the U statistic lies between zero and unity. For two series that are equal, U is zero; the higher the U, the larger the inequality. The coefficient can be further broken down into three components, which indicate the relative contribution of three specific sources to the overall inequality between the two series. This breakdown is particularly interesting, because it shows (1) difference in the means of the series (Um) (i.e., difference in the level of trend growth), (2) difference in the series’ respective variation (Us) (i.e., difference in output volatility), and (3) lack of synchrony or covariation between the series (Uc) (i.e., cyclical asynchrony and changing dynamics of trend growth).

where y¯, x¯, σy and σx are the means and standard deviations of the series y and x, respectively, and ρ is their correlation coefficient.

Figure 3 shows GDP growth inequality as measured by Theil’s inequality coefficients relative to the euro area for three groups over the period 1996 to 2004 Q1: the new member states, the euro area peripheral countries (Greece and Portugal), and the so-called noneuro area EU15 (Denmark, Sweden, and the United Kingdom).8

Figure 3.Theil’s Inequality Coefficients for Growth Between Various Countries and the Euro Area

(Based on year-on-year real GDP growth rates, 1996–2004 Q1)

Sources: Eurostat, national sources, and author’s calculations.

1Excluding Cyprus and Malta.

It turns out that inequality is on average higher for the new member states (mean coefficient 0.49) than for the peripherals (0.30) and the non—euro area EU15 (0.22). Moreover, there has been considerable divergence in inequality among the various new member states. The countries with economic dynamics most similar to the euro area’s are Hungary (coefficient: 0.33), Poland (0.42), and Slovenia (0.45). Meanwhile, Lithuania, Estonia, Latvia, and the Czech Republic post the largest differences.

What explains these inequalities? The statistical components of different growth behavior are shown in Figure 4. As explained above, the statistics reveal how inequality is related to different means, different variances, and a lack of covariance between the individual countries and the euro area. There is a considerable difference in the means and, to a lesser extent, variances of the separate country groups. For the non—euro area EU15, neither means nor variances put economic growth far from that of the euro area. for the peripheral countries, the difference in mean growth is a significant component of inequality. The new member states exhibit an even more sizable mean difference and are, to some extent, subject to a higher lack of covariance, suggesting that their trend growth and cyclical asynchrony amplitudes tend to be larger.

Figure 4.Differences in GDP Growth Compared with Euro Area

(Theil’s inequality coefficients and components, 1996–2004 Q1, based on year-on-year GDP growth rates)

Sources: Eurostat and author’s calculations.

Higher Growth Rates

An obvious feature of the data is that real GDP has expanded considerably faster in the new member states than in the euro area over the past few years. The new member states posted an (unweighted) average GDP growth rate of 4.5 percent from 1996 to the first quarter of 2004, compared with 2 percent in the euro area, 2.5 percent and 3.6 percent in Portugal and Greece, respectively, and from 2 percent to 2.8 percent on average in Denmark, Sweden, and the United Kingdom. Among the new member states, the Czech Republic is a clear outlier, with real GDP expanding by only 1.8 percent on average per year due to the severe banking and stabilization crisis of 1997—99 (Figure 5).

Figure 5.GDP Growth and Standard Deviations in Europe

(GDP, annual percentage change, 1996–2004 Q1)

Sources: Eurostat, and author’s calculations.

Higher growth rates in the new member states compared with the euro area can mainly be explained by the catching-up of these economies, as well as initially the recovery from the “transformational recession” of the first half of the 1990s. While this higher growth is needed to converge with the per capita income levels in the euro area, the accompanying structural differences in economic dynamics may increase the stabilization costs that a new member state would incur if it abandoned its own independent monetary policy. Higher long-term growth rates imply not only higher inflation rates (e.g., through the Balassa-Samuelson effect) but also a risk of inappropriately low nominal, and thus real, short-term interest rates. In combination with a high marginal return on capital, these lower interest rates could fuel a credit boom that, owing to inevitable supply-side constraints facing the investment demands, would give rise to asset bubbles and boom-bust cycles.

Higher Output Fluctuations

In addition to faster economic growth, most new member states have also experienced wider output fluctuations. The average standard deviation of real GDP growth was 2.5 percent in the new member states from 1996 to the first quarter of 2004, higher than in the euro area and the euro area periphery (with 1.2 and 1.4 percent, respectively), although it should be noted that the average in the new member states conceals a broad range, from 1.4 percent in Hungary to 3.7 percent in Lithuania. The five Central European economies—the Czech Republic, Hungary, Poland, the Slovak Republic, and Slovenia—together posted on average a much smaller standard deviation (2.1 percent) than the Baltic countries (3.2 percent). This finding partly reflects the recession in the Baltics in the aftermath of the Russian crisis of 1998 and the subsequent recovery.

Higher output fluctuation in the new member states compared with the euro area can be explained mainly by the fact that the transition process and the implementation of structural reforms have followed a fairly uneven path. Moreover, the new member states have experienced only imperfect access to international capital markets as they have been exposed to stronger changes in investor sentiments. Most important, high investment ratios in most of the new member states, combined with the fact that capital spending tends to be more cyclical than consumption, suggests that during the catching-up period, growth fluctuations will remain larger. Interestingly, the growth differential between the euro area and the new member states has not diminished over the sample period. A period of narrowing growth differences from 1996 to 1999 has given way to a renewed divergence over the past few years (Figure 6). With respect to monetary policy, large differences in output fluctuations could imply that in an enlarged euro area, monetary policy would not be sufficiently countercyclical for countries with higher fluctuations.

Figure 6.GDP Growth

(Quarterly data, annual percentage changes)

Source: Eurostat.

1Weighted by nominal GDP in 2003, excludes Cyprus and Malta.

Lower Synchronization of Business Cycles

While the new member states have on average higher and more volatile growth rates than the euro area, this does not necessarily imply divergent business cycle timing, which is another important factor when discussing whether countries are already well prepared to join the monetary union and to abandon flexible exchange rates. Strongly correlated cyclical swings over time across countries imply that countries are exposed to similar shocks and respond in a similar way.

With regard to the symmetry of economic fluctuations, the assessment is diverse across countries and benchmarks. Furthermore, this parameter needs to be estimated and is thus more subject to judgment. The following analysis again uses data from 1996 until the first quarter of 2004 as well as three different measures to estimate correlations: detrended GDP growth, short-term trends in industrial output growth, and estimated broad cycle components. These approaches have complementary advantages and drawbacks, suggesting that results that hold broadly across methods might be reasonably robust and credible.9

The correlation coefficients of detrended annual GDP growth (at a quarterly frequency) are presented in Figure 7.10 The non-euro area EU15 (Sweden, the United Kingdom, and Denmark) post the highest correlation with the euro area at an (unweighted) average of 0.65. Ireland and Portugal lie in a range of 0.4 to 0.5 (the correlation for Greece being around zero) and the correlation for the new member states is 0.27. Among individual new member states, the high correlation of Hungary and Slovenia with the euro area stands out.

Figure 7.Correlation of Detrended GDP Growth with the Euro Area

(Based on quarterly year-on-year growth rates over 1996–2004 Q1)

Sources: Eurostat, national sources, and author’s calculations.

The disadvantage of using GDP correlation to capture cycle synchrony is that even after long-term trend adjustment, the coefficients may be biased owing to technical correlation. In particular, the correlation of the Central European economies with the euro area may reflect similar weather conditions and calendar factors. A standard way of avoiding the problem is by using filters to extract the short-term GDP trends (through moving averages or medians). However, the available time series data are too short (33 observations) to do this in a meaningful way. Monthly data, which provide more observations, are better suited for such smoothing. They additionally allow one to capture short-term dynamics that may be left unnoticed with lower frequency data (e.g., GDP series) that are available on a quarterly or annual basis.

The most popular proxy for monthly activity is industrial production. Industrial production data are more complete and have a longer history than GDP data. In addition, industry represents a substantial share of GDP in the new member states of Central and Eastern Europe (about 30 percent on average in 2003) and is typically the most decisive sector for cyclical dynamics. This series provides enough observations to extract a short-term trend from monthly growth data using a Hodrick-Prescott filter (with a smoothing factor of 100). Given the higher frequency of the data, this short-term trend is used as a proxy for cyclical fluctuations.

Figure 8 shows correlation coefficients for this measure. As in the GDP analysis, the correlation between the euro area and the three non—euro area EU15 is strongest with an average (unweighted) coefficient of 0.76. It is about 0.52 for the three euro area peripheral economies. The new member states from Central and Eastern Europe post the lowest average correlation coefficient: 0.31. However, the dispersion of the group is very wide. Hungary stands out with a coefficient of 0.93, which is the highest of all countries in the panel. The industry short-term trends of Slovenia, Estonia, and Poland are also relatively well correlated with the euro area. The Slovak Republic and Latvia post some positive correlation in their industry short-term trend with the euro area. Interestingly, unlike in the case of GDP, the Czech Republic is not positively correlated with the euro area, and Lithuania remains negatively correlated.

Figure 8.Correlation of Short-Term Industrial Output Growth Trends with the Euro Area

(Based on monthly year-on-year growth rates over 1996 to mid-2004)

Sources: Eurostat, national sources, and author’s calculations.

As would be expected, industry cycles are more closely aligned than GDP for all country groups. Merchandise trade integration between the euro area and the new member states from Central and Eastern Europe is high, and the bulk of foreign direct investment from West to East is in the manufacturing sector. Finally, manufacturing activity across countries is subject to global cycles, particularly in inventory and investment spending. For all of these reasons, correlation of industry data may overstate the co-movements of the overall economies. And it is the latter that should matter for monetary policy.

Therefore, a broad indicator for the business cycle in the countries examined above has been estimated based on monthly data. Compared with simple GDP growth, this has the advantages of avoiding correlation owing to joined quarterly volatility and of examining dynamics at a higher frequency. In addition, compared with the monthly industrial production series, it incorporates the dynamics of more sectors. Indeed, non-tradables sectors (e.g., retail services, construction) often follow dynamics that are more dependent on idiosyncratic domestic growth factors (e.g., monetary conditions, fiscal policy).

We take three separate monthly indicator sets for each country—the annual growth of industrial production, the annual growth of retail sales volumes, and the annual growth of construction output—and distill from them a joined cyclical factor. In some countries where not all data were available, surveys have been used to capture retail and construction activity.11 The joined cyclical component has been estimated by using a state-space model of the Stock and Watson (1991) type that identifies the cycle as a joined linear component of all sectoral cycles. This component can then be smoothed again by a Hodrick-Prescott filter to rid it of short-term volatility. Correlation coefficients have been computed and are presented in Figure 9. They deliver several important messages.

Figure 9.Correlation of Broad Cycle Trends with the Euro Area1

(Based on monthly year-on-year growth rates over 1996 to mid-2004)

Sources: Eurostat, national sources, and author’s calculations.

1 Excluding Slovenia due to data unavailability.

The (unweighted) average of the correlation coefficients of the Central and Eastern European new member states with the euro area falls substantially, to slightly below zero. The coefficient of the non—euro area EU15 stands at 0.71, while the peripheral euro area countries still show a high correlation of 0.49. This difference relative to industry data is not too surprising, however. Given the geographic proximity, the GDP and industrial production correlations between the euro area and the new member states from Central and Eastern Europe have possibly been biased upward by, for example, joined calendar and weather factors.

To sum up, the correlation of economic fluctuations with the euro area seems to be generally weaker for the new member states from Central and Eastern Europe than for the non—euro area EU15 and the peripheral euro area countries.12 However, at the individual country level, the synchronization of business cycles varies considerably. As for Poland, Slovenia, and Hungary, output fluctuation seems to be rather symmetric with the euro area’s, and cycle correlation is usually within the range of—and sometimes higher than—that of the euro area peripherals, such as Portugal and Greece. Business cycle synchronization with the euro area, however, seems to be lower in the Czech Republic and the Slovak Republic, mainly due to currency turbulence and stabilization episodes in the late 1990s. Lithuania always shows a negative correlation of economic fluctuations with the euro area, which might be explained by rather different economic structures and the relatively low degree of trade integration with the euro area. Results for Estonia and Latvia were not robust across methods.

Concluding Remarks

This paper has examined and illustrated the implications of real convergence—that is, catching-up growth in income and adjustment of the real economic structures toward those prevailing in the euro area—on the patterns of economic dynamics in the new member states. Per capita income levels substantially below those in the euro area and segments of the structure of the real sectors that are still affected by the transition process are among the key economic characteristics of the new member states. While different income levels can be, in principle, compatible within a monetary union, such real convergence may imply differences in economic dynamics—including the level of growth, the magnitude of fluctuations, and the timing of peaks and troughs—that could make a single monetary policy inappropriate for some countries.

The paper, which has highlighted that the question of the link between real convergence and monetary integration is an empirical one, has examined economic dynamics of the new member states and the EU countries over the recent years. It has found that dynamics indeed differ considerably among these countries. In particular, the new member states display higher average growth rate and, overall, a lower covariance—i.e., a lower degree of co-movement—compared with the euro area or other EU members. Although the findings cannot be generalized and indeed support a case-by-case approach, differences in economic dynamics should be looked at carefully when monetary integration strategies are developed, because there may be cases where these differences could call for a more gradual approach. This is likely to be particularly relevant in cases in which monetary policy has been used successfully as a stabilization instrument in the past and in cases where fiscal soundness is weak or where product and labor market flexibility is less advanced so that adjustment to asymmetric shocks could be slow and costly. However, it should also be mentioned that in cases where nominal exchange rates have been stable for a long time, where markets display a satisfactory degree of flexibility, and/or where the fiscal position is sound, differences in economic dynamics would seem more manageable for member states participating in the monetary union.

References

    Backé, Peter, ChristianThimann, OlgaArratibel, OscarCalvo-Gonzalez, ArnaudMehl, and CarolinNerlich, 2004, The Acceding Countries’ Strategies towards ERM II and the Adoption of the Euro: An Analytical Review, ECB Occasional Paper No. 10 (Frankfurt: European Central Bank).

    Corden, W. Max, 1972, “Monetary Integration,”Essays in International Finance No. 93 (Princeton, New Jersey: Princeton University Press).

    De Grauwe, Paul, and YunusAksoy, 1999, “Are Central European Countries Part of the European Optimum Currency Area?” inInclusion of Central European Countries in the European Monetary Union,ed. by Paulde Grauwe and VladimirLavrac (Dordrecht: Kluwer Academic Publishers).

    European Commission, 2002, “Real and Nominal Convergence in the Transition Accession Countries: Selected Issues,” inThe EU Economy: 2002 Review, No. 6 (Brussels: European Commission).

    Pindyck, Robert, and DanielRubinfeld, 1997, Econometric Models and Econometric Forecasts (New York: McGraw-Hill, 4th ed.).

    Stock, James, and MarkWatson, 1991, “A Probability Model of the Coincident Economic Indicators,” inLeading Economic Indicators: New Approaches and Forecasting Records,ed. by KajalLahiri and GeoffreyMoore (New York: Cambridge University Press).

    Süppel, Ralph, 2003, “Comparing Economic Dynamics in the EU and CEE Accession Countries,”ECB Working Paper No. 267 (Frankfurt: European Central Bank).

    Tavlas, George, 1994, “The Theory of Monetary Integration,”Open Economies Review,Vol. 5,No. 2, pp. 21130.

    Other Resources Citing This Publication