Sectoral Busness Cycle Synchronzaton n the European Unon * Antóno Afonso, # $ Davde Furcer + January 2007 Abstract Ths paper analyses sectoral busness cycle synchronzaton n an enlarged European Unon usng annual data for the perod 1980-2005. In partcular, we try to dentfy whch sector for each country s drvng the aggregate output busness cycle synchronzaton. Overall, the sectors that provde the most relevant contrbuton are Industry, Buldng and Constructon, and Agrculture, Fshery and Forestry. In contrast, the Servces sector, the largest one n terms of valued added share, shows a relatve low busness cycle synchronzaton and volatlty, mplyng that t contrbutes only margnally to the aggregate output busness cycle synchronzaton. JEL: E32; F15; F41; F42 Keywords: EMU Enlargement, Stablsaton, Synchronzaton, Sectoral Busness Cycle. * We are grateful to Vto Tanz for helpful suggestons and Renate Dreskena for research assstance. The opnons expressed heren are those of the authors and do not necessarly reflect those of the the ECB or the Eurosystem. # European Central Bank, Drectorate General Economcs, Kaserstraße 29, D-60311 Frankfurt am Man, Germany, emal: antono.afonso@ecb.nt. $ UECE Research Unt on Complexty n Economcs; Department of Economcs, ISEG/TULsbon - Techncal Unversty of Lsbon, R. Mguel Lup 20, 1249-078 Lsbon, Portugal. UECE s supported by FCT (Fundação para a Cênca e a Tecnologa, Portugal), emal: aafonso@seg.utl.pt. + Unversty of Palermo, Department of Economcs, Italy. Unversty of Illnos at Chcago. Address: Department of Economcs (M/C 144), 601 S. Morgan Street, Chcago, 60607, Illnos. Emal:dfurce1@uc.edu, dfurcer@yahoo.t Telephone/Fax: +1-3129962683/ +1-3129963344. The author would lke to thank the Fscal Polces Dvson of the ECB for ts hosptalty.
1. Introducton On 1 May 2004 the European Unon (EU) welcomed ten new members: the Czech Republc, Estona, Cyprus, Latva, Lthuana, Hungary, Malta, Poland, Slovena and Slovaka. In addton, two other countres, Bulgara and Romana, wll jon the EU on January 2007, and other countres are at varous stages of canddacy for membershp n the EU. It s lkely that all these countres wll beneft from jonng n the future the European and Monetary Unon (EMU) n terms of nflaton bas reducton, hgher exchange rate stablty, lower nterest rates, and hgher growth. Therefore, a relevant queston s whether these economes should also expect to have hgh costs from EMU membershp. The theory of the Optmum Currency Area, frst developed by Mundell (1961), and ncludng the classcal contrbutons of McKnnon (1963) and Kenen (1969) stresses the mportance of nternatonal lnkages between the members of the monetary unon to face the loss of the country-ndependent monetary polcy to smooth output fluctuatons. 1 To the extent that the monetary polcy would have contrbuted to stablzaton of cyclcal fluctuatons n the past, the loss of the exchange rate control and of the ndependent monetary polcy can be seen as the stablsaton cost n jonng a monetary unon. Moreover, t has been shown n the lterature that ths stablsaton cost s a decreasng functon of the correlaton between the cyclcal output of the member country and the cyclcal output of the anchor country (n ths case the EMU as a whole). Intutvely, f the busness cycle of a country s very hghly correlated wth the EMU-wde cyclcal output, then countercyclcal monetary polcy conducted by the European Central Bank wll be a very close substtute for the country ndependent monetary polcy. Moreover, busness cycle synchronzaton also has mportant mplcatons for the ablty to mplement a common monetary polcy. The man purpose of ths paper s to analyse sectoral busness cycle synchronzaton n an enlarged European Unon. In partcular, we frst ask whether the busness cycles of the new EU countres are synchronzed and compare them wth those of EMU members, usng annual data for the perod 1980-2005. Second we analyse how busness cycle 1 For some recent contrbutons see Alesna and Barro(2002), Alesna, Barro and Tenreyro (2002), Corsett and Pesent (2002). 2
synchronzaton evolves over tme. Thrd, we try to dentfy whch sector for each country s drvng the aggregate output busness cycle synchronzaton. The results of the paper show that whle for some countres (such as Cyprus, Hungary and Malta) EMU membershp wll be not costly, for the other countres wth negatve or neglgble busness cycle synchronzaton the stablzaton cost could be relevant, at least n the short-run. However, busness cycle synchronzaton seems to have ncreased over tme, suggestng that these costs could be become not relevant n the next future. In terms of sectoral decomposton, the results suggest that, although for each country s possble to dentfy whch sector s able to explan more the aggregate output busness cycle synchronzaton, overall, Industry, Buldng and Constructon, and Agrculture, Fshery and Forestry sectors provde the most relevant contrbuton. On the other hand, the Servces sector depcts relatvely low busness cycle synchronzaton. The remander of the paper s organzed as follows. In Secton Two we present a bref lterature revew of the mportance of busness-cycle synchronzaton, partcularly for the EMU. In Secton Three, we present the emprcal methodology used to evaluate aggregate and sectoral output busness cycle synchronzaton. Secton Four reports the results obtaned, and fnally, Secton Fve summarses the paper s man fndngs. 2. Lterature Revew The lterature on busness-cycle synchronzaton n Europe (and how t compares to the U.S.) s vast. Bayoum and Echengreen (1993) found that demand and supply shocks are more correlated between states n the U.S. than n Europe, and that the U.S. states adjust more quckly to economc fluctuatons than European countres. Usng a dfferent methodology, Wynne and Koo (2000) also found that busness cycles are more algned n the U.S. than n the euro area (of 11 members). Other authors, such as Clark and Shn (1998) and Clark and Van Wncoop (2001), focused on both wthn-country and crosscountry synchronzaton. They found that average wthn-country cyclcal output correlatons are larger than cross-country correlatons, for both the U.S. and European countres (and agan that busness cycles are more synchronzed n the U.S. than n Europe). Recently, Peró (2004), examnng the exstence of asymmetres n ndustral 3
producton n seven European countres for the perod 1957-1998, fnds that several of these countres have algned busness cycles. Other studes have looked at changes n correlaton patterns over tme. Angelon and Dedola (1999) found that the output correlaton between Germany and other European countres has clearly ncreased durng 1993-1997. Fatás (1997), usng annual employment growth rates for regons of France, Germany, Italy, and the UK, found that the average correlaton wth aggregate EU-12 employment growth has ncreased from 1966-1979 to 1979-1992. Furcer and Karras (2006), analyzng cyclcal output for the EU-15 countres found that busness cycle synchronzaton has also ncreased for many countres after the creaton of the EMU. In partcular, ths ncrease n synchronzaton s present n all components of aggregate demand, as well as two supply-sde varables, but t s more pronounced n the trade components (mports and, partcularly, exports). They also showed that the ncrease n trade wthn the EMU area s at least partly responsble for the ncrease n cyclcal synchronzaton. The lterature has also consdered the mplcatons of the EMU for fscal polcy. In fact, unlke other monetary unons, the EMU does not have a central fscal authorty, and stablsaton of shocks s left to the responsblty of the domestc fscal polces of the EMU members. However, the lterature has shown that the ablty of the EMU members natonal fscal polces to smooth shocks s very modest. 2 One mplcaton of ths s that busness-cycle synchronzaton s extremely mportant n the EMU not only because t reduces the probablty of asymmetrc shocks, but also because t makes t plausble to expect the European Central Bank (ECB) to respond to aggregate shocks and to mplement stablsng nterventons wth greater ease. The lterature s much thnner on the analyss of sectoral busness cycle synchronzaton. We beleve that ths s an mportant element and t would provde useful polcy ndcatons, snce t would enable to dentfy whch sector for each country s drvng the aggregate output to be more or less synchronzed wth EMU-wde busness cycle. Thus, the analyss carred out n the followng of ths paper has the purpose to extend the lterature and provde some nsghts n understandng busness cycle synchronzaton. 2 See, for example, Galì and Perott (2003), Afonso and Furcer (2006). 4
3. Emprcal Methodology We obtan the output busness cycle measures by detrendng the seres of real GDP. Four dfferent methods are used to detrend the output seres of each country and estmate ts cyclcal component. The frst measure s smple dfferencng (growth rate of the real GDP). The second and the thrd method use the Hodrck-Prescott (HP) flter, proposed by Hodrck and Prescott (1980). The second method conssts of usng the value recommended by Hodrck and Prescott for annual data for the smoothness parameter ( λ ) equal to 100. The thrd method conssts to consder the smoothness parameter ( λ ) equal to 6.25. In ths way, as ponted out by Ravn and Uhlg (2002), the Hodrck-Prescott flter produces cyclcal components comparable to those obtaned by the Band-Pass flter. The fourth method makes use of the recently popular Band-Pass (BP) flter proposed by Baxter and Kng (1999), and evaluated by Stock and Watson (1999) and Chrstano and Ftzgerald (2003) that also compares ts propertes to those of the HP flter. 3 Whle mnor dfferences among the results obtaned by the three flters are not dffcult to detect (for example, dfferencng generally produces the most volatle seres, whle the BP the smoothest), the man characterstcs are remarkably smlar. Ths robustness wll be formally assessed by the estmatons of the emprcal secton. In practce, we measure GDP busyness cycle synchronzaton for each country as the correlaton between the country s cyclcal component, c, and the EMU s cyclcal component, c EMU : corr ( c, c EMU ). (1) Successvely, n order to dentfy whch sector j for each country s manly responsble for the aggregate output busness cycle synchronzaton, we frst compute the country s sectoral cyclcal components, components and the EMU s cyclcal component: c j, and then we compute the correlaton between these 3 See Appendx 1 for an addtonal descrpon of the flterng methods used n the paper. 5
j ( c c ) corr,. (2) EMU Moreover, to determne the relatve weght of each sector n the computaton of the aggregate output busness cycle synchronzaton, we compute the standard devatons of the country s sectoral cyclcal components. It s clear n fact, that the hgher s the volatlty of a gven sector, the more relevant s ths sector n the computaton of the aggregate output busness cycle synchronzaton. In partcular, f we could approxmate 4 the country s GDP cyclcal component c as the sum of the sectoral value added cyclcal components c c, (3) j j then we could decompose the correlaton between the country s cyclcal component and the EMU s cyclcal component, as a weghted average of the correlatons between the country s sectoral cyclcal components j c and the EMU s cyclcal component: j j ( c c ) w corr( c c ) corr, EMU,. (4) EMU j The weghts, j w, are represented by the share of output busness cycle volatlty (measured by the standard devaton of the cyclcal components) attrbutable to each sector: j w j σ =. (5) σ 4. Emprcal Analyss 4.1. Data We use data from the European Commsson Annual Macro-economc Database (AMECO). 5 Our dataset covers 28 countres (the 12 current EMU countres, the 3 old EU 4 Whch s not possble due to the nonlnearty of the flterng methods. 5 See the Annex for a descrpton of data sources and avalablty. 6
countres whch have not adopted the euro, the 10 new EU members, and 3 prospectve members, Bulgara, Romana and Turkey) from 1980 to 2005. The ncome varable we use to determne output busness cycle synchronzaton s real GDP n 2000 constant prces. We use data for Gross Value Added for the Industry (not ncludng Buldng and Constructon), Agrculture, Forestry and Fshery, Buldng and Constructon, and Servces sectors to decompose output synchronzaton nto sectoral busness cycle synchronzaton. 6 4.2. Output Busness Cycle Synchronzaton We calculate the correlaton coeffcent of each country s cyclcal component of real GDP wth that of EMU, as a whole, usng the HP flter wth smoothness parameter equal to 6.25. Even though the estmated correlatons vary accordng to the detrendng method used, the mpled rankngs are very smlar, regardng the overall perod, the hghest Spearman s rank correlaton coeffcents s 0.936 (BP, HP6.25) and the lowest s 0.776 (Dff, HP100), as can be seen from Table 1. [Insert Table 1 here] Table 2 consders three dfferent perods of analyss. The frst s from 1980 to 1992 and consders the EU15 countres. The second s from 1993 to 2005 and apples to all 28 countres. The thrd s the overall perod from 1980 to 2005. In relaton to the overall perod, we can see that for most EMU countres busness cycle s relatvely well synchronzed, even f for some countres (namely Fnland) there s an almost zero correlaton wth the EMU economy as a whole. [Insert Table 2 here] Lookng at the perod 1993-2005, France shows an almost perfect correlaton wth the EMU economy. However, comparng the 12 eurozed countres wth the 3 (old) non-euro economes, t s dffcult to establsh a systematc relatonshp. In fact, Denmark, Sweden 6 We use the GDP deflator to express the varables n 2000 constant prces. 7
and the UK appear to be more synchronzed wth the EMU-wde cycle than some euro area members, such as Greece and Fnland. The new EU countres show a generally hgher synchronzaton wth the EMU than the canddate countres. In partcular, there are some new EU countres (such as Cyprus, Hungary and Malta) already well synchronzed wth the EMU, and wth correlatons comparable to, or even hgher than, those of some of the old members. On the other hand, several new EU countres (such as Estona, Lthuana and Slovaka) exhbt negatve correlatons, as do two of the three prospectve EU members (Romana and Turkey). The other new EU and accesson countres (namely, Czech Republc, Latva, Poland, Slovena and Bulgara) show a very neglgble, even f postve, correlaton. Overall, we can argue that whle for some countres as Cyprus, Hungary and Malta, EMU membershp wll be not costly, for the other countres wth negatve or neglgble busness cycle synchronzaton the stablsaton cost could be relevant, at least n the short-run. Focusng on the 1980-2005 perod s only fully feasble for the old EU members, but ths can be used to ndcate how the correlatons have changed for these countres, and how they could change for the prospectve Member States. The most strkng fact to emerge from ths exercse s that the degree of synchronzaton wth EMU has remarkably ncreased for all countres (wth the excepton of Germany, where t remaned broadly smlar). Ths result can be largely attrbuted to the achevement of a more ntegrated market snce 1992, and to an ncrease n trade as ponted out by Furcer and Karras (2006). But, perhaps more nterestngly, the results show that the ncreased synchronzaton has been at least as large n the non-euro area as n the euro area economes. Thus, t s lkely to be the case that as the ntra-emu share of trade for the new EU members ncreases, busness cycles wll become more synchronzed. 7 Moreover as ponted out by Frankel and Rose (1998), Rose and Engel (2002), and Rose (2005), the busness cycle synchronzaton between the canddate countres and the currency unon (or the anchor country) s endogenous, and t wll tend to ncrease once the country wll be part of the currency unon. 7 See also Arts and Zhang (1997). 8
4.3. Sectoral Busness Cycle Synchronzaton Industry (excludng Buldng and Constructon) In Table 3, we calculate for each country the correlaton coeffcent between the ndustry value added cyclcal component and that of the EMU-wde GDP, and the standard devaton of the Industry value added cyclcal component (usng the HP flter wth smoothness parameter equal to 6.25 for consstency). For each country, we report also the share of total value added generated by the Industry sector. We also consder n Table 3 three dfferent perods of analyss: the frst s from 1980 to 1992 and consders the EU15 countres: the second s from 1993 to 2005 and apples to all 28 countres; and the thrd s the overall perod from 1980 to 2005. [Insert Table 3 here] Lookng at the shares of ths sector we can see that ndustry generated around one fourth of the total valued added (lookng at the perod where data for all 28 countres are avalable, t ranges from 13.0 percent for Cyprus to 32.4 percent for Ireland). Moreover, comparng the two sub perods t emerges clearly that ths share s dmnshng over tme (n favour of the servce sector as we wll see n the followng of our analyss). In relaton to the overall perod, we can see that for some countres such as France, Germany, Ireland, Italy and Span, the Industry sector synchronzaton wth the EMU-wde GDP s relatvely hgh. Moreover a szeable volatlty of ths sector n these countres contrbutes to provde a sgnfcant contrbuton n the total GDP synchronzaton. In contrast, for other countres such as Austra, Greece and Netherlands, the ndustry valued added cyclcal component s weakly correlated wth the EMU-wde busness, and n the case of Greece t s also worsen by the relatve hgh volatlty. However, lookng at the two sub perods we can see an ncrease n the Industry busness cycle synchronzaton for many EMU countres, contrbutng to the ncrease n GDP busness cycle synchronzaton found n the prevous secton. 9
Fnally analyzng the perod 1993-2005 where the results for all 28 countres are avalable, we can see that whle the new EU and canddate members show a hgher volatlty n the Industry sector, there s no partcular dfference between EU and EMU countres. In fact, whle some of them (such as Cyprus and UK) are more synchronzed than most of the EMU members, other countres (Hungary, Slovaka, Czech Republc and Estona) show a negatve correlaton. Buldng and Constructon In Table 4, we present the results n terms of busness cycle synchronzaton, volatlty and share of total value added for the Buldng and Constructon sector. Lookng at the shares we can see that the valued added contrbuton generated by the Buldng and Constructon sector s relatvely small around 6%. Moreover, comparng the two sub perods t emerges that for many EU15 countres (wth the excepton of Span, Austra and Portugal) ths share s dmnshng over tme. [Insert Table 4 here] In relaton to the overall perod, we can see that for some countres such as Ireland, Belgum and Sweden, the Buldng and Constructon sector s well synchronzaton wth the EMU-wde GDP s relatvely hgh and relatvely volatle, provdng a sgnfcant contrbuton n the total GDP synchronzaton. In contrast, for other countres such as Austra, ths sector s negatvely synchronzed wth the EMU-wde GDP, and for many other countres the correlaton s neglgble. Lookng at the two sub perods we can not observe, as n the case of the Industry sector, a systematc ncrease n busness cycle synchronzaton (wth the excepton of Ireland, Span, the UK, and Portugal). Addtonally, analyzng the perod 1993-2005, we can see that whle the new EU and canddate members show a hgher volatlty n the Buldng and Constructon sector, they have lower busness cycle synchronzaton. In fact, most of them show a negatve or qute neglgble correlaton. Furthermore, t seems that for some countres such as Span and Portugal, characterzed by an ncreasng share and hgh and ncreasng busness cycle synchronzaton n Buldng 10
and Constructon, ths sector contrbutes sgnfcantly and postvely to the computaton of total GDP busness cycle synchronzaton. Agrculture, Fshery and Forestry In Table 5, we present the results n terms of busness cycle synchronzaton, volatlty and share of total value added for the Buldng and Constructon sector. In terms of the share of ths sector n total added value, we can see that t s qute small (wth the excepton of Greece, Portugal, Ireland, and Span), and s decreasng over tme. In relaton to the overall perod, we can see that for most of the EU15 countres the Agrculture, Fshery and Forestry sector synchronzaton wth the EMU-wde GDP s relatvely small. However, t s more than compensated, n terms of relevance n the computaton of the aggregate output busness cycle synchronzaton, by a hgh volatlty (compared to the other sectors). [Insert Table 5 here] Lookng at the two sub perods we can see that whle synchronzaton ncreased for some countres (for example, Austra and Germany) t decreased for many others (especally Ireland). In contrast busness cycle volatlty seems to have remaned qute stable. Fnally analyzng the perod 1993-2005, agan we can see that the new EU and canddate members show a hgher volatlty than n the ndustry sector, but there are no major dfferences between the EU and the EMU countres n terms of busness cycle synchronsaton. Latva seems to be an excepton, beng characterzed by sgnfcant synchronzaton and hgh volatlty. Overall, although ths sector s characterzed by relatvely low busness cycle synchronzaton, t s very volatle, contrbutng nevertheless to GDP busness cycle synchronzaton. 11
Servces In Table 6, we present the same set of results analyzed for the Servces sector. In terms of the share n total valued added, we can observe that the Servces sector s the most relevant sector n the European Unon, around 60 percent, and ts share s ncreasng over tme. [Insert Table 6 here] Analysng the result n terms of GDP busness cycle synchronzaton, n relaton to the overall perod, we can see that for Italy, Span and Sweden the Servces sector s well synchronzaton wth the EMU-wde GDP. For the other countres the correlaton s ether negatve (as n the case of Austra, Netherlands and Denmark) or neglgble. Volatlty s also s relatvely low compared to the other sectors, mplyng a low weght n the contrbuton of the aggregate output busness cycle synchronzaton. Lookng at the two sub perods we can see that whle synchronzaton ncreased for some countres (especally France, and the Netherlands), sgnfcant decreases were recorded for Greece and Italy. For the perod 1992-2005, where the results for all 28 countres are avalable, we can see that the new EU and canddate members show an hgher volatlty n the Servces sector (as n the other three sectors), but agan there s no partcular dfference between EU and EMU countres n terms of sectoral synchronzaton. In fact, whle countres such as Cyprus, Malta, and the UK are more synchronzed than most of the EMU members, other countres (such as Hungary, Slovaka, Czech Republc and Estona) show a negatve correlaton. Overall, t seems that due to the low busness cycle synchronzaton and the relatve low volatlty, ths sector s the one that (compared to the other three sectors) contrbutes less to the aggregate output busness cycle synchronzaton. To conclude our analyss, we can pont out that, although for each country s possble to dentfy whch sector s able to explan more the aggregate output busness cycle synchronzaton, overall, the sectors that provdes the most relevant contrbuton are Industry, Buldng and Constructon, and Agrculture, Fshery and Forestry. In contrast, although the Servces sector s the largest one n terms of valued added share, t shows 12
relatve low busness cycle synchronzaton and volatlty, mplyng that t contrbutes only margnally to the aggregate output busness cycle synchronzaton. 8 5. Concluson The new EU member states are expected to jon n the future the sngle currency. It s lkely, that all these countres wll receve benefts n terms of nflaton bas reducton, hgher exchange rate stablty, lower nterest rates, and hgher growth from jonng the EMU. The theory of the Optmum Currency Area stresses the mportance of busness cycle synchronzaton (the busness-cycle correlaton between the canddate s economy and that of the euro area as a whole) to face the loss of the country-ndependent monetary polcy to smooth output fluctuatons. The results of the paper show that there are some new EU countres (such as Cyprus, Hungary and Malta) already well synchronzed wth the EMU, and wth correlatons comparable to, or even hgher than, those of some of the old members. On the other hand, several new EU countres (such as Estona, Lthuana and Slovaka) exhbt negatve correlatons, as do two of the three prospectve EU members (Romana and Turkey). However, t s worthwhle mentonng that ths analyss can gve useful ndcatons n terms of stablzaton costs only n the short-medum term. In fact, as t has been shown by Frankel and Rose (1998), busness cycle synchronzaton s lkely to ncrease for the EU countres once they would jon the EMU, as EU membershp could ncrease ntra-emu trade allowng busness cycles to become more synchronzed. It s sgnfcant that our analyss also shows that busness cycle synchronzaton has ncreased for all the EU15 countres after the achevement of the Sngle Market (1991). Successvely, we tred to dentfy for each country whch sector s drvng the aggregate output busness cycle synchronzaton. In partcular, we consdered four sectors: Industry (not ncludng Buldng and Constructon); Agrculture, Forestry and Fshery; Buldng and Constructon; and Servces sectors. 8 Addtonally, we report n Appendx 2 results for sectoral busness cycle synchronzaton wthn country, lnked wth the EMU-sector synchronzaton va the EMU-country synchronzaton. 13
For each of ths sector we computed the total valued added shares, the sectoral busness cycle synchronzaton and volatlty. Overall, t seems that whle the Servces sector s the largest one n terms of valued added share, t shows relatve low busness cycle synchronzaton and volatlty, mplyng that t contrbutes only margnally to the aggregate output busness cycle synchronzaton. In contrast, the other three sectors are overall qute synchronzed and relatvely volatle, mplyng a hgher and more relevant contrbuton. Moreover, for each country s possble to dentfy whch sector s able to explan more the aggregate output busness cycle synchronzaton. For example, for countres lke Germany, France, Italy, Cyprus and UK the Industry sector s the one that has the hgher busness cycle synchronzaton. For countres lke Belgum, Span (especally n the last decade) and Sweden the Buldng and Constructon sector s the more relevant. Fnally the Agrculture, Fshery and Forestry sector s partcularly mportant for the Czech Republc and Latva. 14
References Afonso, A and Furcer D. (2006). EMU enlargement, stablzaton cost and nsurance mechansms. Manuscrpt. Alesna, A. and Barro, R. (2002). Currency unons, Quarterly Journal of Economcs 117, 409-436. Alesna, A.; Barro, R. and Tenreyro, S. (2002). Optmal currency areas, NBER Workng Papers, 9072. Angelon, I. and Dedola, L. (1999). From the ERM to the Euro: nee evdence on economc and polcy convergence among EU countres, European Central Bank Workng Paper 4. Arts, M. and Zhang, W. (1997). Internatonal busness cycles and the ERM: Is there a European busness cycle, Internatonal Journal of Fnance and Economcs, 2 (1), 1-16. Bayoum, T. and Echengreen B (1993) Shockng aspects of the European monetary ntegraton, n: Torres F, Gavazz F (Ed.) Adjustment and Growth n the European Monetary Unon. Cambrdge Unversty Press: Cambrdge, 193-229. Baxter, M. and Kng, R. (1999). Measurng Busness Cycles: Approxmate Band-Pass Flters for Economc Tme Seres, Revew of Economcs and Statstcs, 81 (4), 575-593. Chrstano, L. and Ftzgerald, T. (2003). The Band Pass Flter, Internatonal Economc Revew, 4 (2), 435-465. Clark, T. and Van Wncoop, E. (2001). Borders and busness cycles, Journal of Internatonal Economcs, 55, 59-85. Clark, T. and Shn, K. (1998). The sources of fluctuatons wthn and across countres, Federal Reserve Bank of Kansas Cty, Research Workng Paper 4. Corsett, G. and Pesent, P. (2002) Self-valdatng optmum currency areas, NBER Workng Papers, 8783. Fatás, A. (1997). EMU: countres or regons: lessons from the EMS experence, European Economc Revew, 41 (3-5), 743-751. Furcer, D. and Karras, G. (2006). Busness Cycle Synchronzaton n the EMU, Appled Economcs (forthcomng). Galì, J and Perott, R. (2003). Fscal Polcy and Monetary Integraton n Europe, Economc Polcy, 18, 533-572. Frankel, J. and Rose, A. (1998). The Endogenety of the Optmum Currency Area Crtera, Economc Journal 108, 1009-1025. Hodrck, R. and Prescott, E. (1980) Postwar U.S. Busness Cycles: An Emprcal Investgatons, Dscusson Papers 451, Carnege Mellon Unversty. 15
Kenen, R. (1969) The theory of optmum currency area: an eclectc vew, n: R. Mundell and A. Swoboda, (eds.), Monetary problems of the nternatonal economy, Chcago, Unversty of Chcago Press. Kydland, F. and Prescott. E. (1977). Rules Rather Than Dscreton: The Inconsstency of Optmal Plans, Journal of Poltcal Economy, 85, 473-490. McKnnon, R. (1963). Optmum currency areas, Amercan Economc Revew 53, 717-725. Mundell, R. (1961) A theory of optmum currency areas, Amercan Economc Revew 82, 942-963. Peró, A. (2004). Are Busness Cycle Asymmetrc? Some European Evdence, Appled Economcs, 36, 335-342. Ravn, M. and Uhlg, H. (2002). On Adjustng the Hodrck-Prescott Flter for the Frequency of Observatons. Revew of Economcs and Statstcs, 84 (2), 371-380. Rose, A. and Stanley, T. (2005) A meta analyss of the effect of common currences on nternatonal trade, Journal of Economc Surveys, 19 (3), 347-365. Rose, A. and Engel, C. (2002). Currency unons and nternatonal ntegraton, Journal of Money, Credt and Bankng 34, 1067-1089. Stock, J. and Watson, M. (1999). Busness Cycle Fluctuatons n U.S. Macroeconomc Tme Seres, n Taylor, J. and Woodford, M. (eds.), Handbook of Macroeconomcs, vol 1, 3-64, Elsever. Wynne, M. and Koo, J. (2000). Busness cycles under monetary unon: a comparson of the EU and US, Economca, 67, 347-374. 16
Table 1 Spearman s rank correlaton matrx HP6.25 HP100 BP Dff HP6.25 1.000 HP100 0.936 1.000 BP 0.847 0.855 1.000 Dff 0.839 0.776 0.788 1.000 Table 2 Busness cycle synchronsaton (vs-à-vs EMU) 1980-1992 1993-2005 1980-2005 EMU countres Austra 0.534 0.793 0.647 Belgum 0.692 0.832 0.762 Fnland 0.582* 0.478 0.509* France 0.615 0.977 0.786 Germany 0.763 0.678 0.696 Greece 0.601 0.441 0.554 Ireland 0.285 0.645 0.465 Italy 0.539 0.810 0.674 Luxembourg 0.419 0.745 0.570 Netherlands 0.542 0.875 0.692 Portugal 0.341 0.733 0.507 Span 0.506 0.871 0.662 Other EMU Czech Republc 0.031 Denmark 0.043 0.569 0.258 Estona -0.220 Cyprus 0.541 Latva 0.238 Lthuana -0.032 Hungary 0.789 Malta 0.698 Poland 0.247 Slovena 0.412 Slovaka -0.673 Sweden 0.164 0.695 0.443 UK -0.137 0.594 0.042 Canddate countres Bulgara 0.342 Romana -0.242 Turkey -0.273 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 17
Table 3 Industry (excludng Buldng and Constructon) contrbuton to busness cycle synchronsaton (vs-à-vs EMU) Busness Cycle Synchronzaton Volatlty Sector share n the country added value (%) 80-92 93-05 80-05 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra -0.169 0.706 0.284 0.014 0.017 0.015 26.054 22.612 24.402 Belgum 0.174 0.453 0.349 0.025 0.023 0.023 26.333 21.970 24.239 Fnland 0.464* 0.430 0.447* 0.055* 0.061 0.058* 27.361 26.628 27.009 France 0.569 0.815 0.679 0.020 0.016 0.018 22.311 17.704 20.100 Germany 0.501 0.801 0.668 0.017 0.016 0.016 32.747 25.135 29.093 Greece 0.418-0.185 0.175 0.047 0.021 0.036 20.119 14.690 17.513 Ireland 0.527 0.040 32.428 Italy 0.669 0.342 0.547 0.027 0.031 0.030 27.434 23.518 25.554 Luxembourg 0.135 0.592 0.313 0.035 0.028 0.031 13.624 Netherlands -0.001 0.490 0.196 0.023 0.019 0.020 24.459 19.825 22.235 Portugal 0.585 0.233 0.379 0.047 0.027 0.039 22.116 20.033 21.116 Span 0.617 0.442 0.563 0.041 0.030 0.037 26.578 20.909 23.857 Mnmum -0.169-0.185 0.175 0.014 0.016 0.015 20.119 14.690 17.513 Maxmum 0.669 0.815 0.679 0.055 0.061 0.058 32.747 32.428 29.093 Other EU Cyprus 0.656 0.007 13.003 Czech Republc -0.108 0.052 31.315 Denmark 0.021 0.586 0.325 0.025 0.023 0.024 20.831 20.257 20.556 Estona -0.060 0.028 21.908 Hungary -0.193 0.081 26.623 Latva 0.173 0.054 21.002 Lthuana 0.428 0.109 25.239 Malta 0.510 0.049 23.080 Poland 0.009 0.088 25.704 Slovaka -0.122 0.031 28.767 Slovena 0.350 0.080 30.468 Sweden 0.393 0.349 0.374 0.039 0.062 0.051 24.841 24.093 24.482 UK 0.043 0.633 0.303 0.036 0.047 0.041 30.831 22.660 26.909 Mnmum 0.021-0.193 0.025 0.007 0.024 20.831 13.003 20.556 Maxmum 0.393 0.656 0.039 0.109 0.051 30.831 31.315 26.909 Canddate countres Bulgara -0.005 0.384 24.757 Romana -0.070 0.114 30.783 Turkey -0.201 0.124 23.740 Mnmum -0.201 0.114 23.740 Maxmum -0.005 0.34 30.783 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 18
Table 4 Buldng and Constructon contrbuton to busness cycle synchronsaton (vs-à-vs EMU) Busness Cycle Synchronzaton Volatlty Sector share n the country added value (%) 80-92 93-05 80-05 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra -0.295-0.131-0.237 0.026 0.018 0.022 7.235 7.829 7.520 Belgum 0.621 0.602 0.622 0.037 0.024 0.031 5.504 4.944 5.235 Fnland 0.567* 0.407 0.472* 0.107* 0.102 0.105* 7.245 5.114 6.222 France 0.476 0.433 0.477 0.029 0.029 0.029 6.894 5.562 6.255 Germany 0.391 0.088 0.248 0.044 0.034 0.039 5.935 5.571 5.760 Greece 0.231-0.130 0.108 0.056 0.036 0.047 7.583 7.443 7.516 Ireland 0.795 0.041 6.657 Italy 0.782 0.091 0.493 0.043 0.036 0.040 6.520 5.354 5.960 Luxembourg 0.343 0.114 0.218 0.023 0.036 0.029 6.327 Netherlands 0.012 0.765 0.358 0.032 0.023 0.028 5.751 5.463 5.613 Portugal 0.104 0.696 0.152 0.113 0.041 0.085 5.664 6.691 6.157 Span 0.463 0.658 0.516 0.077 0.036 0.062 7.348 8.258 7.785 Mnmum -0.295-0.185 0.175 0.023 0.018 0.015 5.504 4.944 5.235 Maxmum 0.782 0.815 0.679 0.113 0.102 0.058 7.583 8.258 7.7875 Other EU Cyprus -0.639 0.015 7.738 Czech Republc -0.348 0.070 7.474 Denmark -0.171 0.344 0.108 0.066 0.042 0.055 5.461 5.068 5.272 Estona -0.287 0.039 6.056 Hungary 0.016 0.118 4.917 Latva 0.459 0.129 5.292 Lthuana -0.084 0.124 6.888 Malta 0.233 0.024 4.678 Poland 0.339 0.086 6.898 Slovaka -0.511 0.119 5.804 Slovena 0.244 0.090 5.669 Sweden 0.726 0.436 0.601 0.075 0.061 0.071 6.112 4.356 5.269 UK 0.094 0.611 0.296 0.059 0.048 0.054 6.070 5.251 5.677 Mnmum -0.171-0.639 0.108 0.059 0.015 0.054 5.461 4.356 5.269 Maxmum 0.726 0.611 0.601 0.075 0.129 0.071 6.112 7.738 5.677 Canddate countres Bulgara 0.025 0.471 4.442 Romana -0.207 0.124 6.162 Turkey -0.132 0.162 5.325 Mnmum -0.207 0.124 4.442 Maxmum 0.025 0.471 6.162 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 19
Table 5 Agrculture, Fshery and Forestry contrbuton to busness cycle synchronsaton (vs-à-vs EMU) Busness Cycle Synchronzaton Volatlty Sector share n the country added value (%) 80-92 93-05 80-05 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra -0.139 0.546 0.183 0.032 0.024 0.028 4.412 2.335 3.415 Belgum 0.239 0.430 0.290 0.038 0.035 0.036 2.283 1.409 1.863 Fnland 0.635* 0.431 0.507* 0.090* 0.057 0.074* 6.731 3.957 5.399 France -0.023 0.305 0.083 0.035 0.026 0.030 4.176 3.001 3.612 Germany 0.196 0.530 0.295 0.056 0.046 0.050 1.706 1.233 1.479 Greece 0.150-0.179 0.032 0.112 0.042 0.084 12.396 8.054 10.312 Ireland -0.010 0.034 4.697 Italy 0.461 0.040 0.316 0.033 0.035 0.035 4.545 2.970 3.789 Luxembourg 0.001 0.220 0.080 0.049 0.045 0.046 0.812 Netherlands 0.124 0.455 0.236 0.030 0.033 0.031 4.079 2.823 3.476 Portugal -0.071-0.006-0.054 0.047 0.035 0.041 12.871 4.751 8.974 Span 0.197 0.052 0.205 0.053 0.041 0.048 5.926 4.422 5.204 Mnmum -0.139-0.179-0.054 0.032 0.024 0.028 1.706 2.335 1.863 Maxmum 0.635 0.546 0.507 0.112 0.057 0.084 12.871 1.409 10.312 Other EU Cyprus 0.222 0.039 3.992 Czech Republc 0.054 0.054 4.041 Denmark -0.026 0.386 0.180 0.050 0.083 0.066 4.584 2.755 3.706 Estona -0.061 0.029 6.580 Hungary -0.173 0.084 5.209 Latva 0.565 0.111 5.992 Lthuana 0.090 0.131 9.361 Malta 0.113 0.037 2.570 Poland 0.055 0.097 6.146 Slovaka -0.092 0.060 5.136 Slovena 0.251 0.090 3.622 Sweden 0.446 0.188 0.293 0.042 0.058 0.050 3.780 2.189 3.016 UK 0.055 0.073 0.037 0.044 0.028 0.036 1.867 1.298 1.594 Mnmum -0.026-0.173 0.037 0.042 0.028 0.036 1.867 1.298 3.706 Maxmum 0.446 0.386 0.293 0.050 0.131 0.066 4.584 9.361 1.594 Canddate countres Bulgara -0.064 0.270 15.367 Romana -0.307 0.116 15.823 Turkey -0.274 0.165 13.945 Mnmum -0.307 0.116 13.945 Maxmum -0.064 0.270 15.823 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 20
Table 6 Servces contrbuton to busness cycle synchronsaton (vs-à-vs EMU) Busness Cycle Synchronzaton Volatlty Sector share n the country added value (%) 80-92 93-05 80-05 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra -0.338 0.286-0.074 0.017 0.016 0.016 63.614 66.748 65.119 Belgum 0.092 0.370 0.203 0.021 0.014 0.018 65.842 71.675 68.642 Fnland 0.464* 0.303 0.368* 0.041* 0.051 0.046* 59.543 65.516 62.410 France 0.254 0.507 0.360 0.012 0.010 0.011 66.619 73.732 70.034 Germany 0.197 0.170 0.171 0.019 0.015 0.017 59.807 69.222 64.326 Greece 0.226-0.235 0.039 0.049 0.018 0.037 60.036 69.839 64.741 Ireland -0.127 0.029 56.218 Italy 0.787 0.217 0.527 0.026 0.031 0.029 61.502 68.159 64.697 Luxembourg -0.236 0.250 0.012 0.018 0.024 0.020 80.403 Netherlands -0.414 0.444-0.019 0.020 0.014 0.017 65.270 71.813 68.410 Portugal 0.428 0.550 0.311 0.066 0.022 0.049 60.686 68.554 64.463 Span 0.643 0.586 0.576 0.044 0.019 0.036 61.480 62.045 61.752 Mnmum -0.414-0.235-0.074 0.017 0.010 0.011 59.543 56.218 61.752 Maxmum 0.787 0.586 0.576 0.066 0.051 0.049 65.270 80.403 70.034 Other EU Cyprus 0.840 0.015 75.267 Czech Republc -0.235 0.035 57.159 Denmark -0.292 0.152-0.062 0.016 0.008 0.013 69.093 71.921 70.450 Estona -0.102 0.018 60.771 Hungary -0.225 0.106 58.816 Latva 0.520 0.062 66.786 Lthuana 0.432 0.110 58.635 Malta 0.647 0.034 66.745 Poland 0.073 0.077 58.814 Slovaka -0.402 0.049 60.293 Slovena 0.297 0.073 60.225 Sweden 0.797 0.325 0.535 0.041 0.049 0.045 65.268 69.361 67.233 UK 0.340 0.568 0.411 0.034 0.046 0.040 61.231 70.219 65.546 Mnmum -0.292-0.402-0.062 0.016 0.008 0.024 61.231 58.814 65.546 Maxmum 0.797 0.840 0.535 0.041 0.110 0.051 69.093 75.267 70.450 Canddate countres Bulgara 0.027 0.426 55.109 Romana -0.140 0.089 47.233 Turkey -0.196 0.147 56.956 Mnmum -0.196 0.089 47.233 Maxmum 0.027 0.426 55.109 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 21
Appendx 1 Flterng Methods Lettng ( Y ), = ln, the frst measure s smple dfferencng (growth rate of the real GDP): y t, t c. (1), t = y, t y, t 1 The second and the thrd method used n the emprcal analyss make use of the Hodrck- Prescott (HP) flter, proposed by Hodrck and Prescott (1980). The flter decomposes the seres nto a cyclcal ( c, ) and a trend ( ) t t g,, for the smoothness parameter λ > 0 the followng quantty: g, component, by mnmzng wth respect to t T T 1 ( 2 y ) + ( t g, t g, t+ 1 g, t 1 ) t= 1 2, λ. (2) t= 2 The fourth method makes use of the recently very popular Band-Pass (BP) flter proposed by Baxter and Kng (1995), and evaluated by Stock and Watson (1998) and Chrstano and Ftzgerald (1999) that also compares ts propertes to those of the HP flter. The low pass (LP) flter α (L), whch forms the bass for the band pass flter, selects a fnte number of movng average weghts α h to mnmze: K Q = π π 2 δ ( ω) dω, where α ( L) h = α L and h h = K h α K ( ω) = h = α K he ω. K The LP flter uses α K (ω) to approxmate the nfnte MA flter β (ω). Defnng δ ( ω) β ( ω) α( ω), and then mnmzng Q, we mnmze the dscrepancy between the deal LP flter β (ω) and ts fnte representaton α K (ω) at frequency ω. The man objectve of the BP flter as mplemented by Baxter and Kng (1995) s to remove both the hgh frequency and low frequency component of a seres, leavng the busness-cycle frequences. Ths s obtaned by subtractng the weghts of two low pass flters. We defne ω L and ω H, the lower and upper frequences of two low pass flters as respectvely eght and two for annual data. We therefore remove all fluctuatons shorter than two or longer than eght years. The frequency representaton of the band pass weghts becomes α ω ) α ( ω ), and forms the bass of the Baxter-Kng flter, whch provdes an K ( H K L alternatve estmate of the trend and the cyclcal component. 22
Appendx 2 Wthn Country Sectoral Synchronzaton It s easy to show that the sectoral busness cycle synchronzaton wthn country s lnked wth the EMU-sector (j) synchronzaton va the EMU-country () synchronzaton. Indeed, j σ ( cemu, c ) we computedcorr( cemu, c ) = σ σ and j σ ( cemu, c ) corr( cemu, c ) = σ σ, but we can also cemu dsaggregate corr( c, c j ) as follows: EMU c cemu j c j j σ ( cemu, c ) corr( cemu, c ) = (A2.1) j σ( cemu, c ) σ( c, c ) j σ corr( c, c ) σ corr( c, c ) c EMU c whch allows hghlghtng the contrbutons of both corr( c, c j ) and of corr( c, c ) to corr( c, c j ). Therefore, we also computed the wthn country sectoral busness cycle EMU synchronzaton, corr( c, c j ), whch we report below. EMU 23
Table A1 Wthn Country Sectoral Synchronzaton: Industry; Buldng and Constructon Industry Buldng and Constructon 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra 0.611 0.617 0.605 0.352 0.301 0.310 Belgum 0.543 0.707 0.614 0.693 0.599 0.644 Fnland 0.922* 0.742 0.839* 0.769* 0.900 0.813* France 0.193 0.865 0.495 0.172 0.407 0.300 Germany 0.505 0.427 0.453 0.754 0.598 0.680 Greece 0.517 0.271 0.490 0.593 0.121 0.495 Ireland 0.831 0.789 Italy 0.848 0.390 0.618 0.358 0.096 0.271 Luxembourg 0.450 0.334 0.398 0.213 0.236 0.233 Netherlands 0.471 0.214 0.356 0.163 0.592 0.320 Portugal 0.611 0.713 0.647 0.773 0.887 0.760 Span 0.647 0.737 0.702 0.801 0.835 0.805 Mnmum 0.193 0.21 0.36 0.163 0.10 0.23 Maxmum 0.922 0.86 0.84 0.801 0.0 0.81 Other EU Cyprus 0.437-0.276 Czech Republc 0.752 0.442 Denmark 0.837 0.670 0.763 0.743 0.498 0.658 Estona 0.646 0.635 Hungary -0.193-0.012 Latva 0.195 0.223 Lthuana 0.829 0.658 Malta 0.830 0.229 Poland -0.344-0.149 Slovaka 0.323 0.755 Slovena 0.852 0.784 Sweden 0.667 0.841 0.783 0.351 0.688 0.529 UK 0.189 0.536 0.262 0.519 0.576 0.504 Mnmum 0.189-0.344 0.26 0.351-0.276 0.504 Maxmum 0.837 0.852 0.78 0.743 0.784 0.658 Canddate countres Bulgara 0.345 0.370 Romana 0.677 0.640 Turkey 0.668 0.676 Mnmum 0.345 0.370 Maxmum 0.677 0.676 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 24
Table A2 Wthn Country Sectoral Synchronzaton: Agrculture; Servces Agrculture, Fshery and Servces Forestry 80-92 93-05 80-05 80-92 93-05 80-05 EMU countres Austra 0.056 0.218 0.124 0.220 0.199 0.186 Belgum 0.336 0.356 0.346 0.038 0.242 0.107 Fnland 0.719* 0.584 0.682* 0.611* 0.791 0.677* France 0.286 0.412 0.328 0.050 0.432 0.202 Germany -0.004 0.412 0.087 0.696 0.664 0.649 Greece 0.499 0.350 0.483 0.212 0.506 0.253 Ireland -0.089 0.465 Italy 0.186 0.113 0.185 0.570 0.174 0.383 Luxembourg -0.146 0.191 0.009 0.072 0.006 0.037 Netherlands 0.122 0.296 0.223-0.134 0.314 0.018 Portugal 0.442 0.172 0.363 0.691 0.801 0.684 Span 0.664 0.284 0.545 0.532 0.773 0.595 Mnmum -0.146-0.08 0.009-0.134 0.006 0.018 Maxmum 0.719 0.584 0.682 0.696 0.801 0.684 Other EU Cyprus 0.441 0.470 Czech Republc 0.632 0.165 Denmark 0.079 0.459 0.273 0.805 0.342 0.641 Estona 0.627 0.664 Hungary -0.196-0.256 Latva 0.195 0.027 Lthuana 0.887 0.658 Malta 0.160 0.652 Poland -0.338-0.634 Slovaka 0.464 0.514 Slovena 0.628 0.798 Sweden 0.487 0.760 0.669 0.144 0.639 0.458 UK 0.443 0.466 0.441-0.030 0.212 0.045 Mnmum 0.079-0.338 0.273-0.030-0.634 0.641 Maxmum 0.487 0.887 0.669 0.805 0.798 0.045 Canddate countres Bulgara 0.302 0.355 Romana 0.703 0.431 Turkey 0.749 0.716 Mnmum 0.302 0.355 Maxmum 0.749 0.716 Note: Hodrck-Prescott Flter wth smoothness parameter equal to 6.25. * We dd not consder the years 1991 and 1992 to take nto account the Fnland crss n the early 1990s. 25
Annex Data Sources Table B1 Data sources Orgnal seres AMECO codes * Gross domestc product at 2000 market prces - Natonal currency: Data at constant prces. Prce deflator gross domestc product at market prces - Natonal currency; 2000 = 100. Rato: Data at current prces/data at constant prces Gross Value Added at current prces; agrculture, forestry and fshery products Natonal currency: Data at current prces Gross value added at current prces; ndustry excludng buldng and constructon Natonal Currency: Data at current prces Gross value added at current prces; buldng and constructon Natonal Currency: Data at current prces Gross value added at current prces; servces Natonal currency: Data at current prces Note: * seres from the EC AMECO database. 1.1.0.0.OVGD 3.1.0.0.PVGD 1.0.99.0.UVG1 1.0.99.0.UVG2 1.0.99.0.UVG4 1.0.99.0.UVG5 Table B2 Data avalablty GDP I BC AFF S Austra 1980 1980 1980 1980 1980 Belgum 1980 1980 1980 1980 1980 Fnland 1980 1980 1980 1980 1980 France 1980 1980 1980 1980 1980 Germany 1980 1980 1980 1980 1980 Greece 1980 1980 1980 1980 1980 Ireland 1980 1990 1990 1990 1990 Italy 1980 1980 1980 1980 1980 Luxembourg 1980 1985 1985 1985 1985 Netherlands 1980 1980 1980 1980 1980 Portugal 1980 1980 1980 1980 1980 Span 1980 1980 1980 1980 1980 Cyprus 1990 1995 1995 1995 1995 Czech Republc 1990 1990 1990 1990 1990 Denmark 1980 1980 1980 1980 1980 Estona 1993 1993 1993 1993 1993 Hungary 1991 1991 1991 1991 1991 Latva 1990 1992 1992 1992 1992 Lthuana 1990 1992 1992 1992 1992 Malta 1990 1990 1990 1990 1990 Poland 1990 1991 1991 1991 1991 Slovaka 1992 1993 1993 1993 1993 Slovena 1990 1991 1991 1991 1991 Sweden 1980 1980 1980 1980 1980 Unted Kngdom 1980 1980 1980 1980 1980 Bulgara 1991 1996 1996 1996 1996 Romana 1990 1995 1995 1995 1995 Turkey 1980 1980 1980 1980 1980 Note: In the table s reported the frst year where the data s avalable. GDP=gross domestc product; I= gross value added - ndustry excludng buldng and constructon; BC= gross value added - buldng and; AFF= gross value added - agrculture, forestry and fshery products; S= gross value added - servces. 26