Are eurozone economies converging?


The Single Monetary Market in 1993 and successively the advent of the European Monetary Union (EMU) in 1999 constitute two significant institutional changes that should have affected economic activity in member countries.

ByPhilippe d'Arvisenet

This study investigates the extent to which growth rate dispersion has changed during the last few years and whether business cycle synchronisation has increased among eurozone countries. In a currency area, these elements deserve special attention. In a way, they are the two sides of the same coin. While GDP growth dispersion measures the  GDP volatility in a specific period, the synchronisation of business cycles indicates the degree of joint movements in business cycles across countries over a certain period of time. In a currency area, a certain degree of output volatility is likely during a given period of time. Indeed, some countries could be in a catching-up process and therefore post higher inflation and growth rates. However, a protracted period of growth differential is not beneficial for the sustainability of a currency area. While faster-growing economies would need tighter monetary conditions to counter higher inflation rates, other countries would need slower interest rates to stimulate, for instance, their sluggish growth. Different inflation rates would translate the common policy interest rate into lower or even negative real interest rates for the faster growing economies and higher real rates for the other countries. These differences in terms of real interest rates would sustain growth in the more dynamic economies to an even greater extent, while negatively affecting the others, and thus, increase divergences between their cycles.. The main findings of this study can be summarised as follows: – GDP growth dispersion across eurozone countries has remained almost unchanged over the period under consideration (1960-2007), and has decreased only marginally in the last few years. This has been primarily due to significant heterogeneity in trend GDP growth across countries. This element might partially reflect a catching-up process of some economies. By contrast, output gap dispersion has clearly decreased and it is now at an all-time low, which is a clear sign of increasing business cycle synchronisation among eurozone countries. In fact, the correlation throughout the cycle has reached an all-time high, and is now at 0.65. Considering the four largest economies of the area, Germany, France, Italy and Spain, which account for around 80% of total eurozone GDP, the correlation is even higher (around 0.85). – This point is backed by an analysis of lead/lagged correlation across the eurozone countries, seeking to ascertain whether the cycle of a specific country leads the cycles of the other member countries. There is no evidence of any leading country in the zone, as the highest correlation across the largest eurozone economies occurs without any lag. This finding has important consequences in terms of monetary policy, confirming that the ECB should not have a bias in favour of a particular country when setting interest rates. – The export cycle plays an import role in explaining business cycle developments in the major eurozone economies, with the partial exception of Spain. Moreover, as export cycles are highly correlated between countries, apparently trade linkages are one of the main source of the increasing joint movements in business cycles among countries. – With regard to international correlation, there is clear evidence that the US cycle leads the eurozone cycle by around two quarters. Moreover, we found evidence of increasing correlation between the eurozone cycle and those of other industrialised economies, namely, the US, the UK and Japan. This result, combined with an increased correlation of the US economic cycle with that of the global economy excluding the US, does not seem to support the decoupling hypothesis, according to which the world economy and thus the eurozone could decouple from a US slowdown or recession. The article is organised as follows: at the beginning, a brief review of the benefits and costs of a currency area is presented, focusing in particular on the need of country cyclical convergence in a currency area. Then, after describing the methodology used in the article, we analyse GDP growth and output gap dispersion among countries and then the joint movements in the business cycle among them. Lastly, an analysis of the synchronisation of business cycle at an international level is presented. Why we need cyclical convergence in a currency area: benefits and costs of a currency area The theory of currency area was pioneered by Mundell (1961, 1968). In his seminal study, he showed that a currency area might reduce or even eliminate transactional costs, enhance price transparency, stimulate competition, eliminate exchange rate uncertainty and encourage further FDI by maximising the credibility of the exchange rate. Moreover, the creation of a single monetary authority, independent from the various political authorities, might improve price stability and subsequently lead to lower interest rates. In addition, the creation of a currency area might force countries to adopt structural reforms they would otherwise postpone indefinitely. If the member countries of a currency area can no longer use the exchange rate mechanism to increase the competitiveness of their products, they accordingly have to adopt labour and market reforms in order to increase their country’s competitiveness in a larger market with far more competitors. On the other hand, the main cost is clearly the loss of a counter-cyclical instrument of political economy. The need for policy autonomy can be reduced, or eventually avoided, if factor prices are flexible and factor mobility is high. While progress in this sense has been recorded, rigidities in the EMU are still higher than in the other currency area, i.e. the United States. However, if business cycles of the member countries are correlated to some extent and there is not too much heterogeneity in the propagation of shocks, accordingly the cost of losing an independent monetary policy is clearly reduced. If business cycles are correlated and shocks are more symmetric than asymmetric, then it is likely that all the countries of the currency area are in the same cyclical phase; in this case they would need a similar (looser or tighter) monetary policy. Finally, the risks of having asymmetric shocks which could increase cycle divergence, are all the higher the more specialised economies are (see on this point Kenen (1969)). Larger economies with highly diversified industrial structures are less exposed to this risk, compared with smaller and more open economies. Indeed, as reported in the following section of this article, small economies, like Portugal, Finland, Luxembourg and Ireland, recorded higher degree of output gap volatility with respect to larger economies. Various studies have shown that a currency area is an endogenous process (see for instance, Frankel and Rose (1998), Rose (2000), in other words the very creation of a currency area increases country correlation, mainly via an increase in trade linkages between countries. In this case, a demand shock in one country might have a bigger impact on demand for goods of its trading partner, fuelling the propagation of shocks and cycle correlation. Krugman (1993) has challenged this view; as he claimed that the intensification of trade might induce country specialisation, and therefore reduce business cycle synchronisation. Our finding, (see below) is more in favour of the view propounded by Frankel and Rose than Krugman’s. Indeed, correlation among eurozone countries has increased since the creation of the Single Monetary Market in 1993 and the European Monetary Union in 1999. Methodology used In order to gauge the output growth differences among eurozone countries, we have analysed their dispersion of GDP growth and the output gap, measured by the standard deviation in each time period. The standard deviation is computed in both, un-weighed and weighted terms(1). The weights are equal to the share of each country in total eurozone GDP. To separate the series of GDP between its trend and cyclical component, we used the Baxter and King (1995) band-pass filter, which is a typical moving average. This is a classical standard methodology used to break down series. Recent studies on this issues published by the European Central Bank (see, Benal et al 2006) and the European Commission (see Gayer 2007) used this technique in order to separate series into trend and cyclical components The cyclical component is analysed as a share of trend GDP, i.e. the output gap. The definition of the business cycle adopted by Baxter and King is based on Burns and Mitchell (1946), according to whom the cycle covers between 6 and 32 quarters. We follow this assumption which implies that the maximum length of a cycle is 8 years. Therefore, when computing rolling period standard deviation, the length of each period is 8 years. The cyclical component of GDP is particularly useful when seeking to establish whether eurozone economies are synchronised. To measure these joint movements, we computed the correlation of each country’s business cycle with the other EMU member countries. As we considered the EMU with 12 countries(2), there are 66 pair-wise country correlations. To get a synthetic measure of the correlation at each point of time we took the average of these correlations. Then, to check how it has evolved over time we considered an 8-year rolling period correlation. GDP growth and output gap dispersion are mainly based on the AMECO data set from the European Commission, whose data are in annual terms ranging from 1960 until 2009 (forecasts from 2007). The correlation among the 4 largest economies and among their cyclical GDP components is based on Eurostat quarterly data, starting in Q1 1991. Lastly, the correlation across cycles of the largest economies at an international level, such as the eurozone, the UK, the US and Japan, is based on quarterly data drawn from the OECD Main Economic Indicators. The correlation between the US cycle and the cycle of the world economy without the US economy is computed using annual data from the World Bank (constant prices in 2000 US dollars). The cycle’s components are obtained by using the Baxter and King filter. In the section regarding international correlation, we compute a global manufacturing PMI. It is obtained by aggregating the available output index for the manufacturing sectors in the Eurozone, UK, Czech Republic, Russia, Poland, China, India, Japan, Australia, New Zealand and the United States, which accounts for 86% of world GDP, as measured by using World Bank data at constant market prices in 2000 US dollars. The weights correspond to the share of each country in global GDP. The data for the manufacturing index are obtained from NTC Research and from the Institute of Supply Managers (ISM). Output growth volatility between eurozone countries

Heterogeneous country performances during the last years In this section, we analysed the output growth dispersion between eurozone countries. The picture is quite heterogeneous: while all countries experienced a slowdown in GDP growth from 2001 to 2007, country GDP average growth rates are quite different. Since 1996, Spain and other smaller and more open  economies, such as Luxembourg, Ireland, Finland, Portugal and Greece, have recorded GDP growth rates well above the eurozone average, which reflects, to some extent, a catching-up process. Between 1996 and 2000, Luxembourg recorded an average growth rate of 6.2%, Ireland around 10% and Spain more than 4%. Greece, whose average growth rate was only 0.8% in 1992-1995, from 1996 to 2000, recorded an average growth rate of 3.5%. In the same period, France grew 2.8%, in line with the eurozone as a whole, while Germany and Italy grew by around 2%. The same picture is replicated during the last 7 years, with the smaller and more open economies, such as Finland and Ireland and Luxembourg, which expanded at rather higher rates(between 3% and 5%), France was in line with the eurozone (around 2%), and Germany and in particular Italy well below the eurozone average(1.4% and 1.1%, respectively). This element could be related to structural factors, probably reflecting differences in trend output(3). The differences in country GDP growth, which are sometimes rather significant, have also been flagged by a standard deviation analysis, in weighted and un-weighted terms. It has remained relatively stable throughout the period under consideration, and decreased only marginally in the last few years. It was around 1.5 in the 1980s, up to 2.1 in the 1990s and it has now come back to 1.5 (between 2000 and 2007). Eurozone GDP growth is converging towards the lowest country GDP growth rateNevertheless, it is important to stress that the standard deviation provides only a synthetic indicator of growth rate differentials between countries and it could hide other factors that might have affected output growth volatility. While until the mid-1990s, the eurozone’s GDP growth rate was half way between the maximum and minimum country growth rate, since the late 1990s, it has remained very close to the minimum value. To some extent, this reflects the underperforming of some of the largest eurozone economies, as showed before. Confirming this finding, the un-weighted standard deviation of the largest 3 economies is close to weighted GDP growth dispersion for all the eurozone countries. Decomposing GDP into its trend and cyclical components The analysis of dispersion based only on real GDP growth could mask other relevant elements in terms of convergence between eurozone countries. As stated before, countries can experience lower GDP growth because they are facing some structural problem that widens the gap between GDP growth and their potential, or simply because their economic cycle is fluctuating around its potential trend. We have found that GDP growth volatility has remained almost unchanged over the period under consideration. However, breaking the GDP series down into its trend and cyclical components(4) shows hat the relatively stable GDP dispersion was mainly due to the dispersion of trend GDP growth between countries. This could be partially explained by a catching-up process in some countries, which might have a relative higher trend GDP growth with respect to other more mature economies. By contrast, the dispersion of the cyclical component, which is expressed as a share of the trend, i.e. the output gap, is clearly decreasing and has reached an all-time low. The standard deviation of the output gap was around 0.9 in the 1980s and in the 1990s, and as an average over the past few years has decreased to 0.45. This finding clearly shows that business cycles are converging, and this process has accelerated since the beginning of the 1990s as the advent of the European Single Market has boosted integration among countries. Country grouping: the four largest economies versus the others This picture presented above is even far more straightforward considering, on the one hand, the dispersion of trend GDP growth among the largest four economies and the 12 economies on the other hand. The dispersion of the eurozone with 12 countries’ trend GDP growth is far higher than when only the largest economies are considered; this can be due, at least to some extent, to the greater vigour of the smaller economies of the area. Since 2000, however, the gap between the dispersion of trend GDP in the 4 biggest economies and the others has been narrowing. By contrast, when we look at output gap dispersion, we cannot see significant differences between the two groups of countries, supporting the previous conclusion of cyclical convergence. The general picture is that the dispersion of the output gap in both groups is clearly decreasing. Output gap differentials among countries The analysis of countries’ output gap dispersion shows that the small economies of the area have normally experienced a higher degree of volatility. Portugal, Luxemburg, Finland Ireland recorded higher volatility with respect to Italy and France, and the eurozone as a whole. This can be evidence of how larger, and more differentiated, economies tend to show lower output gap volatility with respect to smaller, somewhat less diversified and more open economies that are accordingly affected to a greater extent by external developments. There are, however, some small economies such as Austria, Belgium and the Netherlands, which recorded a degree of output volatility in line with that of the eurozone as whole. The fact that the largest economies recorded a low degree of dispersion (the 4 biggest economies account for about 80% of eurozone GDP) helps explain the low relative level of the eurozone output gap volatility from 1963 until 2006. General picture: output gap volatility is reducing over time All in all, it emerges that output gap dispersion is decreasing among eurozone countries. Splitting the whole period into two parts, we can see that since the advent of the Single Monetary Market, cycle dispersion among countries has progressively reduced. To test whether these results have been affected by the period that was chosen, an eight-year rolling standard deviation for the output gap was computed. Despite some volatility, output gap dispersion, is clearly trending downwards. Business cycle convergence among eurozone countries

GDP growth and business cycle synchronisation

Until this point, we have analysed GDP growth dispersion and output gap volatility in a specific period of time. However, if a currency area is to operate satisfactorily, it is important that all its economies should have a high degree of joint movement in their cycles. Otherwise, the same monetary policy can have positive effects on some countries, but negative effects on the others, increasing divergences among business cycles. Considering an 8-year rolling sample, GDP growth correlation has clearly increased over time. Excluding the small open economies, as they are somewhat less diversified, and hence more subject to output variability, GDP correlation rises further. The correlation among the 4 largest economies, Germany, France, Italy and Spain is indeed relatively high, around 0.8%. As GDP growth also reflects trend developments, it should be more appropriate to compute the correlation of business cycles by considering the output gap rather than GDP growth. After remaining stable during the 1970s and 1980s, the output gap correlation throughout the eurozone, has significantly risen in the last few years, and it is now around 0.65. Considering only the largest economies, business cycle correlation is even higher (more than 0.80), and the degree of correlation is now at an historical high. This confirms what was found in the previous section: convergence among business cycles is increasing and small open economies seem to be characterised by somewhat more idiosyncratic cycles. Indeed the overall correlation is far higher when only the largest economies are taken into account. Country analysis The correlation between each country and its 11 eurozone partners has increased over time for almost all countries of the area. From 1963 to 2007 and only for a sub-period which includes two institutional shocks — the advent of the EMS and of the EMU, France, the Netherlands and Belgium have the highest degree of correlation, while Greece has the lowest. Luxemburg seems to be the only country whose business cycle correlation has decreased over time. By contrast, small economies, such as Ireland and Finland, and to a smaller extent Austria, have recorded a significant increase in their degree of correlation since 1992. Leading and lag correlation among eurozone countries Is there any leading country in the eurozone? This is an important factor in terms of better understanding how the eurozone economy behaves(5). The cross correlation of the German business cycle with that of the three other largest economies at a point of time t = 0, +1,+2….+12, a procedure that we replicated for France, Italy and Spain shows that there is no evidence of a leading country within the eurozone. All countries have the highest correlation at time t = 0, i.e. a contemporaneous correlation, which reinforces the finding that eurozone business cycles are highly correlated. In terms of monetary policy, this confirms that the ECB should not have any bias for a particular economy when setting interest rates. Not even Germany, which accounts for around 33% of total eurozone GDP, could pass this test(6).. Which component leads growth in each eurozone country? It is useful to analyse which GDP component drives an upturn in every country and how these components are correlated among countries. In this case, we can obtain a better and clearer picture of business cycle synchronisation. The upturns considered to gauge changes in demand components started in Q1 93 and in Q2 03. Both set of data represent the trough of a cycle, with GDP growth contracting in these periods. In both cases, 8 quarters after the trough have been taken into account. During the first expansion phase, export momentum among countries has been rather heterogynous, contrary to what was observed in the second period. One important difference concerns German exports which clearly outperformed the other countries in terms of export growth since early 2003. This development can be, at least partially, explained by the improvement in German price competitiveness versus its eurozone partners(7). While the momentum of investment was rather similar in the first phase, this does not seem to have been the case during the second expansion phase. In particular, in the second period, Germany’s investment momentum has been sluggish and Italy performed only relatively better. Similar conclusions can be drawn from the analysis of consumption. While consumption growth was particularly strong in the 4 largest economies during the first expansion phase, this does not seem to have been the case in the second phase. In particular, the Italian and German performances have been particularly poor with respect to those posted by France and Spain. Once we have analysed which component has proven to be the driving force during the upturn, it is useful to study the correlation of the cyclical part of each GDP component with GDP, in order to understand which component might have a prominent role in explaining movements in the business cycle of each individual country(8). Germany reported high cotemporaneous correlation with exports (around 0.8) and somewhat lower with consumption and investment, although they are rather high (around 0.7). The business cycle in France seems to be more related with the investment cycle (more than 0.9). Nevertheless, the consumption and export cycles are relatively highly correlated with the GDP cycle (the contemporaneous correlations are around 0.7). In Italy the business cycle has the highest correlation with the export cycle (around 0.8), while the correlation with the consumption cycle is quite poor (only 0.3). By contrast, the Spanish business cycle shows a weak correlation with export cycles, and a relatively high correlation with consumption (more than 0.9).. The bottom line is that the export cycle, with the exception of Spain, plays an important role in explaining business cycle developments within the major economies. A rolling period analysis  shows that the correlation of export cycles across these economies has clearly increased, highlighting the fact that an increase in trade linkages is one of the main factors accounting for joint movements in business cycles among countries. Are Central and Eastern European countries correlated with the eurozone? According to the Treaty establishing the European Union (protocol on the third stage of Economic and Monetary Union), all countries belonging to the Union, with the exception of the United Kingdom and Denmark(9), will eventually adopt the euro. The Stability and Growth Pact set the criteria in terms of inflation, public finance, interest rates and exchange rate that a country has to meet before joining the eurozone. For the time being, after the last European Union wave of enlargement was concluded in January 2007, only three of the 12 new member countries have joined the eurozone; Slovenia on 1st January 2007 and Cyprus and Malt on 1st January 2008. Slovakia should adopt the euro on 1st January 2009 and Romania in 2014. Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, and Poland do not currently have a target deadline for joining the eurozone. As stated above, the cost of losing an independent monetary authority is reduced if business cycles are related. The correlation between the cycle of each Central and East European country and the eurozone cycle over tow subperiods. The first started around the mid-1990s and ended in 2004, while the second considered only the periods 2000-2004, helps investigating whether during the last few years the efforts undertaken by these countries to join the EU have increased their correlation with eurozone countries. Slovakia, although it is expected to join the eurozone next year, seems to have a very poor correlation with the eurozone cycle. Romania seems to be in a better position, with a stable correlation of around 0.4. Regarding the other countries, the Czech Republic has a relatively high correlation with the eurozone (around 0.9) while Poland has seen its correlation increase over time. Considering the period between 2000 and 2004 the correlation is around 0.7. With regard to the Baltic countries, only Latvia seems to have a business cycle that is relatively well correlated with the eurozone; lastly, Hungary and Bulgaria have seen their correlation decrease over time. It should be kept in mind that all these countries are relatively young market economies and are currently in a catching-up process, which might imply a period of lower correlation with respect to more mature economies. Moreover, the analysis should be interpreted with precaution given the very short period available(10). International business cycle synchronisation The correlation of business cycles among major economies has become a crucial issue recently. The possibility that the Asian economies will decouple from the US cycle, might imply that the Asian countries could be less affected by a recession or, at the best, a sharply deceleration in the US economy. This might have important consequences for the eurozone economies, as it might imply a lower correlation between the US and the eurozone cycles. Who leads the eurozone cycle? An analysis of lag/lead correlation, considering the period (1973q1 2004q4), shows that the US business cycle leads the eurozone cycle by around two quarters. If the contemporaneous correlation is rather high (0.54), after two quarters it rises to 0.65, and only marginally decreases to 0.59 after three quarters. The correlation between the UK and the eurozone business cycle  shows a similar trend. The highest correlation is observed after two quarters, i.e. around 0.60. Interestingly, although the UK is the major trade partner of the eurozone (exports shipped to the UK account for more than around 15% of total exports outside the eurozone, while exports to the US account for around 13% of total exports outside the eurozone) the euro area’s cycle has a higher correlation with the United States. This element shows that trade linkages, although important, are not the only channel through which shocks spread from the US to the eurozone and cycles are related. Financial and confidence spillovers are increasingly significant transmission channels of shocks. According to our estimates, a 1% decline in US GDP growth will reduce eurozone by around 0.4-0.5% after one year, while considering only the trade link the effect should be close to 0.3%(11). Lastly, there is no clear evidence that the Japanese cycle leads the eurozone cycle. The correlation at contemporaneous lags and with one or two lags are almost identical and relatively low with respect to the correlation with the US or UK cycle. By contrast, the highest correlation between the US and the UK cycle seems to occur without lags, which testifies how the two economies are interdependent. This element has also been confirmed empirically; the Granger causality test showed that there is no clear evidence as to which cycles causes the other. In both cases the null hypothesis that the US does not cause the UK cycle and that the UK cycle does not cause the US cycle have been rejected (see Table). By contrast, dealing with the US-EMU cycles, the hypothesis that the US does not cause the EMU cycle is rejected, but the hypothesis that the EMU cycle does not cause the US cycle is accepted, confirming that the US leads the eurozone cycle. Lastly, with respect to the UK and EMU, causality goes in both directions, implying that there is no clear evidence as to which cycle leads the other. How has the correlation has changed over time between the UK, the US and the Eurozone ?, an 8-year rolling period analysis shows that US-EMU correlation and the UK-EMU correlation have a similar path. During the first half of the 1980s the correlation was rather high, then towards the late 1980s it decreased, due mainly to the financial disruption which affected the US and the UK to a greater extent than the EMU, before subsequently shrinking once more during the early 1990s, at the same time as Germany’s reunification. Since the late 1990s, the correlation has been rather high, reflecting both the increasing correlation of business cycles within the eurozone as well as the increased financial market integration, confidence spillover and trade linkages at an international level(12).  By contrast, the correlation between the UK and US cycles has remained more stable over the period considered, reflecting the closer links between the two Anglo-Saxon countries.

Evidence of decoupling? s there evidence of a “world business cycle”? To verify this hypothesis, we analysed the correlation between the cycle of the US, UK, Japan and Eurozone, whose GDP accounts for around 70% of the world’s GDP. The rolling correlation of their business cycles shows that the correlation not only is increasing, but it is up to an all-time high. This finding contrasts somewhat with the hypothesis that the eurozone can decouple from the US cycle. If the latter was less related to the former, thanks to the robust growth of Asian economies, then we should have observed a lower correlation. Given the weigh of the US in the global economy(13) and, more importantly, the extent of financial integration in the world economy, we are currently seeing shocks which seem, partly at least, symmetrical rather than simple asymmetric shocks(14). Moreover, as the correlation between the US cycle and the rest of the world’s cycle has increased (15), the extent to which shocks spread from one area to another has also risen. A similar conclusion can be drawn from survey data. In the manufacturing sector, back in October 2007, there was a rather significant divergence between the global PMI and the PMI excluding the most advanced economies, namely the Eurozone, the US, the UK and Japan, which pointed to a potentially positive impact of Asian countries, notably China and India, on industrialised countries. In recent months, this gap has narrowed significantly, challenging the hypothesis that emerging countries could decouple from a US recession or severe slowdown. According to the IMF (2007), around 70% of countries have experienced a slowdown in activity once the US economy slipped into recession. Moreover, given that the share of emerging Asia’s exports (Eastern and Southeastern Asia) shipped to the rest of the world is about 70% (Monetary Authority of Singapore 2007) of total exports, compared with 30% shipped to emerging Asia, a slowdown in domestic demand in developed economies would lead to a significant slowdown in export growth for Asian countries. If we add that exports account for a large share of GDP, (in China it is around 38%, in Korea 40%, in Taiwan around 70%), then it is likely that a slowdown in developed economies leads to a significant slowdown in Asian GDP. To sum up, there is evidence of an increasing output gap correlation among the largest economic areas. A shock which could affect one of these economies is likely to spread rapidly to the other advanced areas. Given the high share of total exports shipped from emerging countries to advanced areas, the Asian economies do not seem sheltered from a slowdown in domestic demand in advanced economies. Therefore, the hypothesis that Asia might decouple from the US and then that the eurozone might be more isolated from a US recession or, at best, a sharp deceleration, thanks to the vigour of Asian economies does not seem a plausible hypothesis. Considering the impact of an US slowdown on eurozone countries, Broyer and Brunner (2008) show that Ireland and Belgium (small but open economies) would be significantly affected, cumulating the effects of financial and trade linkage. With regard to the European Union, the UK will suffer from a US slowdown, mainly due to the financial linkage, while the Czech Republic will be affected mainly through the trade channel. Conclusions This article has looked into GDP growth dispersion and business cycle synchronisation among eurozone economies. There is evidence that GDP growth dispersion has remained roughly unchanged over the period considered (1960-2007), and only marginally decreased in the last few years. This seems to be mainly due to a relatively high dispersion of trend GDP growth among countries, which could be related, at least to some extent, to a catching-up process in some countries. However, it is also true that countries, like Germany but in particular Italy, have underperformed the average of the eurozone as a whole in the last few years. This finding is likely to mask, particularly in the case of Italy, the need for structural reforms to boost growth. By contrast, output gap dispersion is clearly trending downwards, pointing to an increase in business cycle synchronisation among eurozone countries. This finding has been confirmed by analysing the joint movements of eurozone busyness. Cycle correlation has considerably increased since the advent of the European Single Market in the early 1990s and the European Monetary Union in 1999. This is an important factor with regard to the correct operation of a currency area. Indeed, if business cycles are synchronised, the cost of having a single currency for different countries is reduced. Risks that the same policy interest rate might induce increasing divergences among country business cycles are definitely moderate. One point deserves to be explored more in depth in a future study; there is evidence of a still high dispersion of trend GDP growth among eurozone countries. It would be worthwhile to analyse the determinants of trend GDP growth through a production function approach, and see what the main differences between countries are. This study could provide a useful guide in terms of understanding differences in terms of GDP growth between countries and what are the prospects for each country and for the eurozone as a whole.


(1) Both methods have their justifications. If it is worthy to analysis GDP growth dispersion per se, as an empirical analysis, it is also true that the European Monetary Union, is a “weighted average phenomenon”. The HICP for the eurozone, is a weighted average, and the ECB’s 2% ceiling target for inflation in the medium term to maintain price stability is set on the HICP. Basten (2006), considering weighted average GDP growth, found evidence of a significant increase in GDP growth correlation in 1999-05 with respect to previous periods.(2) As the series for Malta, Cyprus and Slovenia are quite short, we focused on the Eurozone with 12 countries, that is Austria, Belgium, France, Italy, Germany, Greece, Netherlands, Finland, Spain, Portugal, Luxembourg and Ireland.(3) For a discussion of structural reforms concerning the three largest economies of the eurozone see F. Cerisier, P. Ciocca, S. Costagli, M. Kaiser, E. Vergnaud: “How is the Lisbon strategy faring? Progress of structural reforms”, Conjoncture, BNP Paribas, January-February 2008.(4) See previous section for a description of the methodology used.(5) As the 4 largest economies accounts 80% of the eurozone GDP and because long series in quarterly data for GDP and GDP component for the other countries are not available, the remaining part of this section focuses only on Germany, France, Italy and Spain.(6) This finding supports the conclusion drawn by De Lucia and Lucas, who, estimating an interest rate reaction function for the ECB, did not find evidence of a German bias. De Lucia C., Lucas J.M. (2007) “How different are the Fed and ECB?”, Conjoncture, BNP Paribas, April 2007.(7) De Lucia C. “Eurozone: Is the euro to blame for all price-competitiveness problems?” EcoWeek 08-03, BNP Paribas, January 2008.(8) The cyclical components of consumption, export and investment have been obtained using the Baxter and King (1995) filter. See the methodology section for details(9) The UK and Denmark benefit from the opting out clause, according to which they might decide not to take part in the third stage of economic and monetary union (EMU) and therefore not adopt the euro.(10) Eurostat data for Hungary, Lithuania, Poland and the Czech Republic are available from Q1 1995; data for Bulgaria are available since Q1 1994 and from Q1 1993 for Estonia. By contrast Latvia and Slovakia have a longer data set which start in Q1 1990. As the Baxter and King procedure to break down a series into its trend and cyclical component is subject to the truncation problem (the first and last three years of observation are lost), accordingly there are few data available to compute the correlation.(11) De Lucia C. “ECB: An interest rate hike in December, but what afterwards?” EcoWeek 06-44, BNP Paribas, November 2006.(12) Kaiser obtained similar results, computing an international correlation based on industrial production data. Kaiser M. “Eurozone Uncompleted Convergence” Conjoncture, BNP Paribas September 2005.(13) According to the IMF (2007), the weight of the US in the global economy has increased, accounting for about 30% of global GDP at market exchange rates, roughly 20% of imports and around 50% of stock market capitalisation.(14) d’Arvisenet Ph. “Did you say decoupling?”, Economic Market Monitor, BNP Paribas, January-February 2008.(15) The negative correlation recorded in the late 1990s is probably due to the Asian crisis. Dées and Vansteenkiste (2007) found that the US cycle and the world economy ex the US also had a positive correlation in 1998-1999, when the Asian countries hit by that financial crisis are excluded.