DNB WORKING PAPER Prie onvergene in Europe from a maro perspeive: Trends and deerminans (1960-2003) Riemer P. Faber and Ad C.J. Sokman No. 12/Oober 2004
Prie onvergene in Europe from a maro perspeive: Trends and deerminans (1960-2003) Riemer P. Faber and Ad C.J. Sokman *Views epressed are hose of he individual auhors and do no neessaril refle offiial posiions of De Nederlandshe Bank. Working Paper No. 012/2004 Oober 2004 De Nederlandshe Bank NV P.O. Bo 98 1000 AB AMSTERDAM The Neherlands
Prie onvergene in Europe from a maro perspeive: Trends and deerminans (1960-2003) B Riemer P. Faber * and Ad C. J. Sokman ** Absra This paper sudies he effes of more han 40 ears of European inegraion on pries. Up o now, mos empirial researh in his area has been miro-based. We follow a maro approah. On he basis of saled HICP srong evidene is found for prie onvergene in Europe, espeiall in he 1960s and 1990s. For he DM zone onvergene is found in all deades unil he German re-unifiaion. Ne o os onvergene, rade and a harmonizaion, we demonsrae ha ehange rae sabili has been an imporan faor behind his proess. Ke words: Eonomi inegraion, Inernaional prie dispersions, Prie onvergene, Law of One Prie, EMU, DM zone JEL Classifiaions: E31, E50, F15, F40 * Universi of Groningen. This paper is wrien during an inernship a De Nederlandshe Bank. ** De Nederlandshe Bank, Researh Division. Views epressed are hose of he individual auhors and do no neessaril refle offiial posiions of De Nederlandshe Bank. We would like o hank all pariipans in he DNB lunh seminar of Augus 18 h 2004 and Maro Hoeberihs and Le Hoogduin for helpful advie.
2 1. Inroduion The las five deades, European ounries made a huge effor o inegrae heir naional markes. The foundaion of he European Union (EU), he Single Marke Trea and reenl he inroduion of he euro have been milesones in he proess owards eonomi, monear and poliial unifiaion in Europe. As an approimaion for inegraion, one migh ake prie dispersions beween European ounries sine more inegraed and ompeiive markes are epeed o lower prie differenes. This makes Europe an ineresing es-ase for a sud o he law of one prie (LOP) and is friions under inensive marke inegraion poli. Prie onvergene in Eonomi Monear Union (EMU) has imporan impliaions for he ondu of he European Cenral Bank s (ECB) monear poli as well (Rogers, 2001). Differen naional inflaion raes and dnamis ompliae he seing of he ineres rae whih is based on prie developmens for he euro area as a whole. Finall, eperiene wih prie onvergene in EMU an help predi fuure prie developmens of new EU member saes (Maier, Cavelaars, 2004). The wo main objeives of his paper are (1) o esablish he een pries in presen EMU-member ounries have onverged sine 1960 and (2) o idenif he main deerminans behind his proess. Conrar o oher empirial sudies ha ake a miro individual-produ perspeive our analsis is based on saled Harmonised Indies of Consumer Pries (HICP). B saling we mean ha we ransform HICP indies ino omparable figures on he basis of oasional measuremens b he European Commission of relaive (absolue) HICP levels for individual ounries. Seion 2 onains a brief overview of he miro-evidene of prie onvergene in Europe. In Seion 3 we presen new evidene based on saled HICP daa for differen groups of European ounries. In Seion 4 we desribe a small heoreial model summarizing he poenial auses of prie onvergene. This heoreial eposé provides us wih he ingrediens of a more formal saisial analsis of prie onvergene in Seion 5 for wo disin eonomi regions: he ounries belonging o EMU and he DM zone. Finall, Seion 6 summarizes our main findings. 2. Eising evidene of prie onvergene in Europe and auses of LOP failure In an influenial lieraure surve Rogoff (1996) onludes ha inernaional good markes are highl inegraed, bu no e nearl as inegraed as naional good markes. Large deviaions from he LOP eis and are quie persisen (see for eample Isard, 1977). Engel and Rogers (1996) show evidene for a border effe. This is he par of inernaional prie dispersions ha an no be eplained b phsial disane. Border effes should deline if inernaional markes beome furher inegraed. However, mos sudies suffer from a lak of daa. Prie daa onsrued and published b offiial saisial agenies are indies, whih make a dire omparison of absolue prie levels beween
3 ounries impossible. Onl inflaion rends in eonomies an be dedued from hese indies (see for eample Cehei, Mark and Sonora, 2002 and for Europe Bek and Weber, 2001). A presen, hree pes of daa have been used in prie onvergene analsis for Europe. Firs of all, here are daa on individual produs. There are man papers ha sud individual produs like ars, hamburgers - he well known BigMa inde - magazines or furniure (e.g. Goldberg and Verboven, 2004, Luz, 2002, Haskel and Wolf, 2001). These sudies provide a mied and fragmened sor. Some indiae srong onvergene, some a bi and some find no onvergene a all. Beause of he limied sope, hese sudies are no ver useful for poli making. A seond daa soure is onsrued b UBS (2003) onaining pries of 114 individual goods and servies in 70 iies. Daa are olleed ever hree ears sine 1979. On he basis of his se UBS finds prie onvergene in EU in 1990-1994. The mos ofen ied daa se used in analzing prie onvergene is provided b he Eonomis Inelligene Uni (EUI). This se onains 160 goods and servies from 128 iies worldwide and is olleed annuall sine 1990. On he basis of his informaion, Engel and Rogers (2004), Rogers (2001, 2002) and Rogers e al. (2001) all provide evidene of European onvergene in he earl 1990s. Causes for he failure of he LOP are summed up b Rogoff (1996). He menions radiional eplanaions like ransporaion oss, ariffs and non-ariff barriers like language, ulural differenes, differen loal produ requiremens, inomplee informaion e. Also shor run demand faors an influene prie dispersions (Froo, Rogoff, 1995). Parsle and Wei (2001) find indeed ha disane, uni-shipping oss and ehange rae variabili eplain prie differenes beween iies in he Unied Saes and Japan. Moreover, Parsle and Wei (2002) argue ha for iies all over he world ehange rae sabili lowers prie dispersions and ha formal long erm agreemens have a larger impa han shorer, less formal agreemens. Finall, also priing o marke an be a possible ause of prie dispersions (Krugman, 1987). In ha ase arbirage is no possible or is oss are oo high o make i profiable. Rogers (2002) alulaes orrelaions beween European prie developmens and possible auses of onvergene in he 1990s. He finds ha a and rade liberalisaion, ehange rae sabili and inome onvergene are epla naor faors, wih he firs wo being more imporan han he las wo. 3. New evidene of prie onvergene in Europe based on saled HICP Daa Beause of he limiaions of miro-based researh, we aemp o apure long-erm developmens of prie dispersion in Europe b using aggregaed daa on onsumer pries. Sandard HICPs are aken as
4 basi prie daa. The nie feaure of he HICP is ha i ensures omparabili of onsumer pries aross ounries, and ha daa are available for a long period. 1 Moreover, in onras o earlier menioned miro daa HICP s are represenaive for a omplee ounr insead of onl one large i, daa is olleed monhl insead of annuall, far more produs are used and ofen several observaions are used o ompose one final produ prie. Besides ha, HICPs show eonomi signifian prie differenes beween ounries faed b onsumers (EC, 2002). 2 Major disadvanage of he HICP is ha i is an inde. However, for our analsis of prie onvergene absolue prie levels are indispensable. To make he ounr indies omparable in absolue erms he are resaled on he basis of measured relaive absolue HICP levels for he ear 1999, whih were made publi b Eurosa (see Table 1 below). Absolue prie levels for oher ears are derived bak and forh on he basis of he developmen of he HICP-inde. Furhermore, sine all original daa are repored in loal urren, he HICPs are onvered ino a ommon urren for whih we ook he German mark/euro. The hoie of he numéraire does no affe he size of our dispersion measure (see Eq 2 below). The resuling saled HICPs are shown in Figure A1 of Anne A. Measure of prie dispersion For our analsis we need a measure of prie dispersion. Suppose we have saled he HICPs ounries belonging o group C in ear. The spread of pries in ear ( σ b he sandard deviaion: C p p for n ) an now be alulaed (1) σ p n p p wih p n p C 2 = 1/ ( ), = 1/ C C To make a meaningful omparison hrough ime level p. This is our prie dispersion measure, shorl C σ p needs o be adjused for he average prie C p (Eq. 2). I is a number beween 0 and 1. Wih full prie onvergene, p = p for all, he dispersion measure equals zero. σ p p = = n p p C C 2 1/ (( / ) 1) (2) p C 1 Soure: OECD Eonomi Oulook (Number 75, June 2004). Unless menioned oherwise, all daa is obained via his soure. 2 Remember ha HICPs are olleed o measure prie level hanges of aual onsumed goods. Samples refle aual onsumpion paerns sine individual produ aegories are weighed via differen naional weighs. Samples and weighs also slighl hange wih hanging onsumpion paerns hroughou he ears. Consumer prie indies are ommonl-used o ompare prie hanges (see for eample Cehei, Mark and Sonora, 1998 and Engels and Rogers, 1996).
5 Table 1 Relaive HICP levels 1999 (de=100) Relaive HICP 1999 Ausria (a) 94.2 Belgium (bg) 96.8 Swizerland (h) 130.0 German (de) 100.0 Denmark (dk) 114.9 Spain (es) 74.4 Finland (fi) 108.0 Frane (fr) 93.9 Unied Kingdom (uk) 99.2 Greee (gr) 73.3 Ireland (ir) 95.0 Ial (i) 83.8 Luembourg (lu) 97.4 The Neherlands (nl) 91.7 Norwa (no) 115.1 Porugal (p) 66.6 Sweden (se) 109.5 Soure: Eurosa Chronos Prie dispersion in Europe In Figure 1, he dispersion measure C p is ploed for several ombinaions of ounries. The DM zone group onains ounries ha had alread linked heir urren o he German mark a an earl sage. The EMU group and EU group onain respeivel all EMU and EU members in 2003. The Europe group omprises of all EU members plus Norwa and Swizerland. The preise group omposiions an be found in Table A1 of Appendi A. The DM zone and he EMU-group are aken as our saring poin for a deailed saisial analsis in Seion 5. The figure below displas a delining rend in prie dispersion for all groups. The proess of prie onvergene has been sronges in he DM zone boh in absolue and in relaive erms. From 1960 on HICP onsumer prie levels seadil onverged in he DM zone. In he seond half of he 1980s prie dispersion reahed is lowes level lose o zero. A he beginning of he 1990s prise dispersion in he DM zone rose somewha, possibl as a resul of German re-unifiaion. In more reen ears, prie dispersions delined again. Prie differenials wihin he EMU and EU have alwas been subsaniall higher. Compared o he DM zone, prie onvergene a he European level ook plae a a lower rae. Espeiall in he 1970s and 1980s, he proess of prie onvergene - ha haraerized he 1960s - sagnaed ompleel. Ineresingl, European prie dispersion levels dereased subsaniall in he 1990s. The Europe group has he highes dispersions. Noie ha here is a differene beween EU members and non-eu members sine 1992. Compared o 40 ears ago, prie dispersion in Europe has been halved.
6 Figure 1 Prie dispersions 1960-2003 0.3 0.25 0.2 dp 0.15 0.1 0.05 0 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 Year Europe EU EMU DM zone Is presen level is omparable o prie dispersion levels in he DM zone in he earl siies. In Seion 5, we will disuss in greaer deail he reasons behind hese observaions. To gain insigh in wha happened wihin he large EU group, we inrodue wo new groups: he firs omprises of norhern EU ounries and he of seond souhern EU ounries. Figure 2 shows ha norhern ounries have smaller prie dispersions wih he DM zone group han souhern ounries. Moreover, if we ombine he norhern wih he souhern ounries levels of prie dispersion are he highes. Dispersion in all ounr groups is delining. These ouomes are no surprising. Figure 2 Prie onvergene in Norhern and Souhern pars of he EU 0.4 0.35 0.3 0.25 dp 0.2 0.15 0.1 0.05 0 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 Year DM zone / EU norh DM zone / EU souh EU norh / EU souh DM zone
7 Ineresingl, he overall piure ha emerges from our maro-approah ompares surprisingl well wih he miro-evidene of he previous seion. In Figure 3, a omparison is made wih Rogers (2002) showing he same downward rend in dispersion for he period 1990-2001. 3 Bu also he sagnaion in he 1980s aording o he UBS daa are in aordane wih our dispersion measure (see Figure A2 in Anne A). To sum up: saled HICPs form a promising daase for analzing prie onvergene developmens in eonomi regions. No onl does his approah represen a broad range of produs, i also provides a enaive window daing bak o 1960. A he same ime, our resuls sugges ha miro panel daa like EUIs migh be a good firs approimaion for prie developmens a he onsumer prie level. Figure 3 HICP vs Rogers (2002) rend 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 ear HICP EMU Rogers (2002) EMU 4. A heoreial framework for prie dispersion Whih faors ause prie dispersions? In his seion several eplaining variables will be disussed. There are 2 ounries, X and Y, ha boh produe 2 produs, A wih onsumer prie onsumer prie p a and B wih p b. Produ A (B) from ounr X is eal he same produ as A (B) from ounr Y. A is a radable produ and B is a non-radable produ. There is full ompeiion and here are no profis. Assume furher ha boh produs need a manufaurer and a reailer. The manufaurer sells his produ o he reailer and he reailer sells he produ o he onsumer. Assumed is, following Engel and Rogers (1996), ha one he produ is in he sore i is no radable anmore. So produ A is onl radable when i omes direl from he manufaurer. 3 The Rogers (2002) EMU line is he average of Rogers (2002) CPI weighed raded goods prie dispersions and his CPI weighed non-radeables prie dispersions.
8 The manufaurer makes produion oss (m) and he reailer selling oss (r). Eamples of selling oss are markeing oss, he oss of having a shop e. For boh m and r apial and labour are required as inpu. There is full mobili of labour and apial beween produs and ompanies, bu no mobili of labour beween ounries. Capial is mobile beween ounries. These assumpions are he same as in he Balassa-Samuelson model (Balassa, 1964, Samuelson, 1964). Furhermore assume ha labour is heaper in ounr Y, while apial - due o is mobili - has he same prie in boh ounries. So manufaurers and reailers in ounr Y an produe heaper han manufaurers and reailers in ounr X: m > m and r > r. Tradable produs (A) The prie of produ A produed and sold in ounr X ( he oss of he reailer and he naional a rae: p a ) is deermined b manufauring oss, p a = [ m + r ] * (1 + a ) (3) Similarl, he prie of produ A manufaured and sold in ounr Y is: p a = [ m + r ] * (1 + a ) (4) Sine m > m i beomes profiable for reailers in ounr X o bu produ A from a manufaurer in ounr Y. Noe ha reailers in he EU do no have o pa aes over inpu faors, even when he are impored from anoher EU ounr. 4 The prie of produ A produed in ounr Y for sale in ounr X is deermined aording o Eq.5. Manufauring oss are made in ounr Y and reailer oss in ounr X. Besides loal oss, he reailer has o pa for imporing he produ o ounr X. These arbirage oss onsis of ransporaion oss 5, ehange rae oss (oss of onvering urren and hedging ehange rae risk) and oss o overome rade resriions (Rogoff, 1996). Trade resriions ould be for eample impor ariffs, imperfe informaion or non-ariff barriers ha are osl and are passed on o he onsumer prie. Finall, he naional a rae is harged over he impor prie, loal oss and arbirage oss. p a = [ m + r + Arbirage oss] * (1 + a ) (5) 4 A he naional level aes have o be paid over added value; reailers an reover paid aes on inpu variables. In he EU epors are ze ro raed and aes are levied where he good is onsumed (Eijffinger, De Haan, 2000). 5 These are in equaion (3) inluded in he loal oss sine ransporaion is in ha ase ompleel done in ounr X
9 Arbirage oss = ranspor oss + ehange rae oss + rade resriions (6) These equaions give some insigh in he proess of arbirage. If p a < p a, arbirage is profiable. If ( m - m ) is lose o zero, arbirage oss have o be low o make imporing profiable. The same pe of argumen is made b Krugman (1987). Arbirage is onl profiable if prie differenials beween ounries are relaivel large. If arbirage oss are high, markes are segmened and priingo-marke is possible. In ha ase he manufaurer of produ A ould also earn a profi. To inegrae markes, arbirage oss have o be lowered. If arbirage ours, due o ounr X s impors of produ A here will be era demand for A and less demand for A. As a resul, labour beomes less sare in ounr X and sarer in ounr Y. Consequenl, produ A beomes more epensive and produ A heaper. Arbirage will our unil p a = p a. This is he equilibrium. This does no mean ha p a and p a r 6 and a raes onverge and if arbirage oss deline prie differenes beome smaller. are he same. If Non-radable produs (B) The prie of produ B in ounr X ( p = [ m + b r ] * (1 + p b ) is deermined b a ) (7) And for ounr Y: p = [ m + b r ] * (1 + a ) (8) Remember ha m > m and r > r. Now, boh loal and produion oss are speifi o he ounr where he produ is sold. Produion and loal oss sill onain a apial and labour par. Prie differenes an our due o differenes in produion and loal oss and due o a rae differenes. Aording o he Balassa-Samuelson effe prie onvergene of non-radable goods will our. Sine demand for produ A will inrease due o higher epors, more labour will be reruied for he radable seor. This auses a derease in labour suppl for he non-radable seor. Wages will rise, whih will ause m and r o rise and pb o rise as well. As a resul, pb onverges o p b. To summarize: Prie differenials beween ounries an arise due o faors like arbirage oss, differenes in produion oss, loal oss and aaion. 6 In his ase r and r an onl differ due o differen labour oss. If apial is no full mobile, differen pries for apial ould also ause produion oss differenes.
10 The model an easil be eended o monopolisi ompeiion. In ha ase differen profi margins also eplain inernaional prie differenes. 5. Saisial anals is In he previous seion we ook a brief look a he faors ha are relevan in he one of prie onvergene beween ounries. In his seion a more formal saisial analsis is performed for he EMU-11 and he DM-4 zone. 7 HICP prie-dispersion will be eplained b ransformaions of he following basi variables (see Anne B for a preise definiion): - Following Rogers (2002), inome per apia of a ounr (shorl ) is aken as a measure of produion oss. To isolae ehange rae disurbanes and a effes, GDP per head is epressed in PPP dollars and faor oss. Counr differenes for a given ear are summarized b he dispersion rae; - As a measure of a ounr s a level, he value added a rae is aken (shorl τ ). Counr differenes are alulaed b our dispersion measure; - Arbirage oss are spli up. The ehange rae par is measured b he ehange rae volaili (evol) viz. a viz. he DM/euro. The average aross ounries is he measure we use in our regression analsis. The oher par, rade resriions and ransporaion oss, is almos impossible o measure (Prihe, 1996). Here we follow he lieraure (see for eample Dollar and Kraa, 2002) and ake openness of a region (open), measured b inra ross-border goods rade as % region s gdp, as a pro; - Business le posiions (bus) of ounries migh have an impa on prie differenials, for eample via he mark-ups whih are learl lial in Europe (Griffih and Harisson, 2004). The sandard deviaion of real GDP growh aross ounries is he relevan measure. The rends of he EMU-11 and DM-4 zone variables and prie dispersions are depied in Figure 4. 8 Differenes in a raes and inome wihin he regions and ehange rae volaili have diminished subsaniall hrough ime. A he same ime, ounries wihin EMU have beome muh more open. In our regression analsis, we will eplore o wha een hese developmens have onribued o lower prie dispersion in EMU and he DM zone. 7 We have eluded Luemburg from our analsis. As he dispersion measure is no ounr weighed, Luemburg would have had an disproporionae influene on he overall resuls. 8 GDP per apia (faor, in PPP) for he DM-4 zone is orreed for a large downward jump afer he German reunifiaion.
11 Figure 4 Eplanaor variables DM-4 zone and EMU-11 (1960-2003) Dispersion HICP Dispersion Indire a 0.3 0.12 0.25 0.1 0.2 0.08 dp 0.15 dt 0.06 0.1 0.04 0.05 0.02 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 ear ear Dispersion GDP per apia (faor, in PPP) Sandard deviaion real GDP growh 0.4 6 0.35 0.3 5 0.25 4 dy 0.2 0.15 0.1 0.05 obus 3 2 1 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 ear 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 ear Openness Ehange rae volaili (4 ear mean) 40 2.5 35 30 2 OPEN 25 20 15 10 5 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 ear evol4 1.5 1 0.5 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 ear DM-4 zone EMU-11 Of ourse, we have o keep in mind ha a omplee deomposiion of prie dispersion ino a number of driving faors is no possible. For eample, lower ehange rae volaili has a dire posiive impa on prie onvergene, bu also an indire one as i simulaes inernaional rade and hereb openness (see for eample Rose 2000, Sokman, 1995). One needs a large sale muli-ounr model o separae he differen influenes, whih is beond he sope of his paper. Wih he eepion of he sandard deviaion of bus, all variables are firs-order non-saionar (see Table 2). For his reason, we adop an error-orreion speifiaion for our regression analsis.
12 In shor: p = α + ασbus β ( p β β τ β open β evol β β dum) (9) 0 1 0 1 2 3 4 5 6 1 + + + - + +/- wih p : dispersion of saled hip σ bus : sandard deviaion gdp growh : dispersion gdp per head (faor oss, in PPP) τ : dispersion value added a raes open : epors wihin ounr group as a share of gdp evol : -ear moving average ehange rae volaili : rend (1, 2, 3,.) dumm : 1 from 1992 onwards (Single Marke Trea) dume : 1 from 1999 onwards (Inroduion euro) dum91/93 : 1 for he ears 1991, 1992 and 1993 Noe ha he firs par of Eq. (9) eplains he dnami developmen and he seond par he long erm relaion. The residuals of he individual long erm equaion do no onain an uni roo for boh he EMU-11 and he DM-4 zone, onsisen wih oinegraion. Also inluded are a ime rend (1, 2, 3, ) and hree dumm variables ha ake he value 1 afer he Single Marke Trea (dumm, 1992), he inroduion of he euro (dume, 1999) and he hree ears afer he German re-unifiaion (1991-93). From Seion 4 follows ha lower a rae dispersions and os differenials, more snhronized business les and lower ehange rae oss and volaili lower prie dispersions. Furhermore, i is epeed ha openness simulaes prie onvergene. Table 2 Augmened Dike-Fuller uni roo ess EMU-11 DM-4 zone 1964-2003 X? (X) X? (X) p -0.82 * -4.81-2.51 * -2.56 ** -2.11 * -6.53-1.28 * -5.48 evol -2.70 * -12.22-2.47 * -7.49 τ -2.81 * -4.07-2.71 * -4.69 open -1.54 * -7.37-2.56 * -7.43 σ bus -4.60-6.23 * Uni roo a 95% onfidene level ** Uni roo a 90% onfidene level
13 Esimaion resuls Equaion (9) has been esimaed for he eonomies belonging o EMU-11 and he DM-4 zone and over he period ranging from 1965 o 2003. The main resuls of our OLS regression analsis are summarized in Table 3. Ne o he eplanaor variables lised in Eq. (9), we have eperimened wih oher poeniall relevan variables, like spli-ups of GDP in a labour and apial inome par, oil pries, oher proies for openness and so on. These hanges did no signifianl improve he resuls. Furhermore, some of he variables are epeed o direl influene pries, like oss and aes 9. Ohers ma ake some ime, like ehange rae volaili. Parsle and Wei (2002) find ha long erm ehange rae sabili lowers prie dispersions more han shorer erm ehange rae sabili. Prie onvergene is epeed if people rus ha ehange raes are sable for a longer period. Therefore, ehange rae volaili is esed for one ear and 4 or 8 ear moving averages. 10 The lef half of Table 3 onains he regression resuls for he EMU-11 and he righ half he ones for he DM-4 zone. The resuls are promising. For he EMU-11 we find in he shor run ha prie dispersion is signifianl affeed b differenes in business le ondiions aross ounries. 11 In he longer erm, ne o he impued a dispersion, inpu oss, openness and ehange rae volaili onribue and are signifian wih he epeed signs. Noie ha ehange raes have o be sable for a long period before i has a posiive effe on prie dispersion: he eplanaor power of he 8 ear moving average is muh higher han he 4 ear average of ehange rae volaili (Column 1). This an eplain wh he EMU dumm is no signifian. A ime rend improves he resuls, jus like a highl signifian Maasrih dumm (Column 3). This suggess ha he Maasrih Trea mean a rue break in peoples and insiuions prospes onerning he European inegraion proess. The lowes lines of Table 3 presen some sabili and auo orrelaion ess whih are all passed. The olumns 4 o 6 show similar resuls for he DM-4 zone ounries. Again prie onvergene an be eplained b long erm oss onvergene, inreased openness and long erm ehange rae volaili. Noie ha he 4 ear volaili measure performs beer in he DM-4 zone han he 8 ear measure (Column 1). This makes sense, as he DM-4 zone had eperiened a muh longer period of ehange rae sabili han oher European ounries. This also migh eplain wh we ould no esablish an 9 The onribuion of a dis persion o prie dispersion is deermined (in a ehnial sense) b he share of aes in he prie. In our regression analsis we, herefore, fi he aompaning oeffiien β 2 in Eq (9) a he average a rae of he ounr group. This resriion ould no be rejeed b he Wald-es. 10 Moreover, all variables from he long erm par were also esed in he dnami par. This did no improve he resuls. 11 The 1990 value of σbus our findings. is high for boh EM U-11 and he DM-4 zone. Dropping hese values does no affe
14 Table 3 Summar esimaion resuls Equaion (9) EMU-11 (1) (2) (3) 1965-2003 DM-4 zone (4) (5) (6) Eplanaor variables a) σ bus 0.005 (2.0) 0.006 (2.4) 0.005 (2.3) 0.001 (1.5) 0.002 (2.3) 0.002 (2.2) β 0 0.68 (5.5) 0.69 (5.9) 0.75 (6.7) 0.23 (3.2) 0.24 (3.8) 0.23 (3.8) 1 τ τ 1 open 1 1.56 (7.8) 1.51 (8.7) 1.27 (7.2) -0.16 (0.4) 0.39 (1.4) 0.76 (1.9) 1 ( - ) b) 1 ( - ) 1 ( - ) 1 ( - ) ) 1 ( - ) 1 ( - ) -0.003 (0.6) -0.003 (0.8) -0.010 (2.1) -0.007 (3.5) -0.006 (4.0) -0.006 (3.9) 0.006 (0.6) - ( - ) - ( - ) 0.048 (1.8) 0.039 (2.7) 0.046 (2.9) evol4 1 0.049 (3.2) 0.055 (4.7) 0.051 (5.2) -0.042 (1.0) - - evol8 1 0.004 (3.7) 0.004 (3.8) 0.006 (4.8) 0.003 (4.1) 0.003 (4.9) 0.003 (4.2) dumm - - -0.032 (2.4) - - - dum91/93 - - - - 0.037 (2.4) 0.035 (2.3) dume - - - - - -0.025 (1.5) 2 R 0.48 0.49 0.56 0.33 0.48 0.51 Q 1 0.6 <0.44> 0.7 <0.41> 1.9 <0.17> 0.0 <0.97> 0.7 <0.40> 1.1 <0.30> Q 2 2.0 <0.38> 1.9 <0.39> 2.3 <0.31> 0.4 <0.83> 1.0 <0.61> 1.4 <0.50> ChowB 1985 16.5 <0.04> 8.4 <0.30> 6.5 <0.59> 21.0 <0.01> 11.6 <0.17> 8.8 <0.45> ChowF 1990 28.1 <0.01> 23.3 <0.06> 16.6 <0.28> 23.8 <0.05> 13.0 <0.53> 9.4 <0.81> 1992 26.1 <0.01> 20.5 <0.06> 13.9 <0.31> 11.6 <0.48> 9.0 <0.71> 5.4 <0.94> 1999 5.9 <0.31> 3.6 <0.61> 4.2 <0.53> 3.7 <0.60> 4.8 <0.44> 1.2 <0.94> Noe: Beween ( ) are -values, beween < > P-values a) Eplanaor variables are zone-speifi (EMU-11 or DM -4 zone) b) Wald-es oefτ τ =1 has a P-value of 0.33 1 ) Wald-es oefτ τ =1 has a P-value of 0.43 1 auonomous Maasrih-effe for he DM-4 zone. The German reunifiaion also had a signifian impa (signifian dum9193) on prie differenials. The proess of prie onvergene in he DM-4 zone worsened for a ouple ears. The eogenous ime-rend is also signifian and has an upward slope, herefore parl offseing he downward enden in one or wo oher faors sine i is unepeed ha i refles an unidenified enden owards inreased prie dispersions. Again a he boom of he able some ess are presened whih are all passed. Wihin he group of EMU-11 ounries, here are wo disin periods wih rapidl falling prie dispersion: he 1960s and he 1990s and furher. Our saisial analsis shows ha in boh periods several faors have been working in he same direion. During boh episodes, we winessed
15 onverging GDP-levels, harmonisaion of aes, ehange-rae sabili and growing inernal rade (see also Figure 4). In he period in beween, riggered b he ollapse of Breon Woods, he proess of prie onvergene sagnaed. Larger prie dispersions were driven b growing differenials in os levels and high levels of ehange rae risk. In he period 1989-2003, prie dispersion levels sared o deline again whih aording o our invesigaion was aused b narrowing os differenes and falling ehange rae volaili, along wih growing openness and a onvergene. Also Maasrih had a posiive impa on onvergene. In he DM-4 zone, prie onvergene had been a gradual proess, onl emporaril disruped b he German reunifiaion. Prie dispersion levels in he DM-4 zone have alwas been well below he levels for he EMU-11 area as a whole. German and he hree small neighbouring ounries have reahed a high level of inegraion in erms of pries, os sruures, aes, ehange raes and bilaeral rade. 6. Conlusion This paper shows he effes of more han 40 ears of European inegraion on pries. Via saled HICPs srong evidene for prie onvergene in Europe is found, espeiall in he 1960s and 1990s. For he DM zone onvergene is found in all deades unil he German reunifiaion. Saisial analsis for he EMU and DM zone shows ha a harmonisaion, rade liberalisaion, oss onvergene and ehange rae sabili are eplaining faors for deviaions of he LOP and driving faors behind European prie onvergene. Espeiall oss onvergene and ehange rae sabili seem o be imporan. There is a rend of onvergene o he LOP whih is aused b delining friions. The resuls sugges ha EU sueeded in is goal o inegrae naional markes. However, prie dispersions wihin EMU are a his momen higher han wihin he old DM zone suggesing furher prie onvergene is possible. This argumen is fed b he large effe of long-erm ehange rae sabili in EMU. However, European prie dispersions are alread below he US level (Rogers, 2001). Up o now prie dispersions sine he sar of EMU dereased no subsaniall in omparison wih earlier ears. This suggess ha is onsequenes for monear poli are limied ill now. Based on he impa of EU membership, he Single Marke Trea in pariular, one an sa ha prie onvergene an be epeed for ounries joining he EU.
16 Appendi A Table A1 Group omposiions Europe EU EMU EMU-11 DM zone DM-4 zone EU norh EU souh Ausria (a) Belgium (bg) Swizerland (h) German (de) Denmark (dk) Spain (es) Finland (fi) Frane (fr) Unied Kingdom (uk) Greee (gr) Ireland (ir) Ial (i) Luembourg (lu) The Neherlands (nl) Norwa (no) Porugal (p) Sweden (se) Figure A1 Saled HICP indies in Dm-euro 1960-2003 Prie level 160 140 120 100 80 60 40 20 0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 Year a bg h de dk es fi fr uk gr ir i lu nl no p se
17 Figure A2 HICP vs UBS (2003) (rend) rend 1979 1982 1985 1988 1991 1994 1997 2000 2003 ear HICP EMU UBS (2003) EMU
18 Appendi B Variable definiions Taes The a rae on HICP produs, he indire a rae, for ounr in ear ( τ ) is alulaed aording o he following formula. Indire Taes τ = (b1) Privae Consumpion The dispersion in naional a raes is measured on he basis of 4 ears moving averages, sine indire a levels var wih business le movemens and he average lengh of a business le is abou 4 ears. Produion and Loal oss GDP per apia for ounr is measured agains faor oss and in PPP dollars as follows: ( GDP Indire Taes + Subsidies )/ PPP Toal Populaion = (b2) For ever ear, he sandard deviaion is aken for ounr group C ( σ C σ is divided b he mean of C for ounries in group C: C ). To make a omparison in ime C σ = (b3) Business le posiion We ake real GDP growh as a pro for he onjunural phase ounr is in ( bus ). Again he sandard C deviaion is aken over he ounries under onsideraion ( σ bus ). Trade Resriions As a measure oal inra-eu epor as perenage of GDP of EU ounries is aken for he EMU-11 and a GDP weighed openness for he DM4 zone ( open ). 12 C Ehange Raes To measure annual ehange rae sabili for ounr, he monhl volaili of he naional urren wih respe o he German mark is alulaed ( evol ). 13 The following formula is used (EC, 1990): evol 2 ( E E ) = (b4) m 12 m E m is he monhl perenage hange of he ehange rae of he urren of ounr for monh m wih respe o he German mark-euro and ounries from group C is alulaed ( evol ). E is he average of C E m in ear. For all evol of ear he mean of 12 Soure: European Commission; European Eonom, The EU eonom: 2002 review No 6 / 2002, and The EU eonom: 2003 review No 6 / 2003, saisial anne 13 Soure: 1960-2003 OECD Olis, 1957-1959 Inernaional Finanial Saisis d-rom (IMF, Sepember 2004).
19 Referenes Balassa, B. (1964), The purhasing power pari dorine: a re-appraisal, Journal of Poliial Eonom, 72, 584-596 Bek, G. and Weber, A. (2003), How wide are European borders? New evidene on inegraion effes of monear unions, Cener for Finanial Sudies Working Paper No. 2001/7 Cehei, S., Mark, N., and Sonora, R. (2002), Prie inde onvergene among Unied Saes iies, Inernaional Eonomi Review, 43(4), 1081-1099 Dollar, D., Kraa, A. (2003), Insiuions, rade, and growh, Journal of Monear Eonomis, 50, 1, 133-162 Eijffinger, S. C. W. and De Haan, J. (2000), European monear and fisal poli, Oford Universi Press, Oford Engel, C. and Rogers, J. H. (1996), How wide is he border?, Amerian Eonomi Review, 86, 1112-1125 Engel, C. and Rogers, J. H. (2004), Euro s prie dispersion, Eonomi Poli, Jul, 347-384 European Commission (1990), One marke, one mone, European Eonom, 71-74 European Commission (2002), European inegraion and he funioning of produ markes, European Eonom, Speial repor No 2 / 2002 Froo, K. A. and Rogoff, K. (1995), Perspeives on PPP and long-run real ehange raes, in G. M. Grossman and K. Rogoff (eds), Handbook of Inernaional Eonomis, volume III, Elsevier Siene, Norh-Holland, 1647-1688 Goldberg, P. K. and Verboven, F. (2004), Marke inegraion and onvergene o he Law of One Prie: evidene from he European ar marke, Journal of Inernaional Eonomis, forhoming Griffih, R. and Harisson, R. (2004), The link beween produ marke reform and maro-eonomi performane, European Commission Eonomi Papers No. 209 Haskel, J. and Wolf, H. (2001), The Law of One Prie A ase sud, Naional Bureau of Eonomi Researh Working Paper No. 8112 Isard, P. (1977), How far an we push he Law of One Prie?, Amerian Eonomi Review, 67, 5, 942-948 Krugman, P. R. (1987), Priing o marke when he ehange rae hanges, in S. W. Arnd and J. D. Rihardson (eds), Real-finanial linkages among open eonomies, MIT Press, Cambridge MA, 49-70 Maier, P. and Cavelaars, P. (2004), Convergene of prie levels: lessons from he German reunifiaion, Journal of Comparaive Eonomis, forhoming Luz, M. (2002), Prie onvergene under EMU? Firs esimaes, Working Paper, Insiue of Eonomis, Universi of S. Gallen
20 Parsle, D. C. and Wei, S. -J. (2001), Eplaining he border effe: he role of ehange rae variabili, shipping oss, and geograph, Journal of Inernaional Eonomis, 55, 87-105 Parsle, D. C. and Wei, S. -J. (2002), Limiing urren volaili o simulae goods marke inegraion: a prie based approah, Vanderbil Universi and Inernaional Monear Fund Prihe, L. (1996), Measuring ouward orienaion in LDCs: an i be done?, Journal of Developmen Eonomis, 49, 307-335 Rogers, J. H., Hufbauer, G. and Wada, E. (2001), Prie level onvergene and inflaion in Europe, Working Paper 01-1, Insiue for Inernaional Eonomis, Washingon D.C. Rogers, J.H. (2001), Prie level onvergene, relaive pries and inflaion in Europe, Inernaional Finane Disussion Paper No. 2001-699, Board of Governors of he Federal Reserve Ssem, Washingon D.C. Rogers, J. H. (2002) Monear union, prie level onvergene, and inflaion: How lose is Europe o he Unied Saes?, Inernaional Finane Disussion Paper No. 2002-740, Board of Governors of he Federal Reserve Ssem, Washingon D.C. Rogoff, K. (1996), The purhasing power pari puzzle, Journal of Eonomi Lieraure, 34, 647-668 Rose, A. K. (2000), One mone, one marke: he effe of ommon urrenies on rade, Eonomi Poli, 30, 7-33 Samuelson, P. (1964), Theoreial noes on rade problems, Review of Eonomis and Saisis, 64, 145-154 Sokman, A. C. J. (1995), Effe of ehange rae risk on inra-ec rade, De Eonomis, 143, 41-54 UBS (2003), Priing and earnings around he globe, Zurih
Previous DNB Working Papers in 2004 No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 No. 7 No. 8 No. 9 No. 10 No. 11 Jaob A. Bikker, Laura Spierdijk and Pieer Jelle van der Sluis, Marke Impa Coss of Insiuional Equi Trades J.W.B. Bos and C.J.M. Kool, Bank Effiien: The Role of Bank Sraeg and Loal Marke ondiions Maro Hoeberihs, Mewael Tesfaselassie and Slveser Eijffinger, Cenral Bank Communiaion and Oupu Olivier Roodenburg, On he prediabili of GDP daa revisions in he Neherlands Iman van Lelveld and Franka van Liedorp, Inerbank Conagion in he Duh Banking Seor Joke Mooij, Corporae Culure of Cenral Banks: Lessons from he Pas David-Jan Jansen and Jakob de Haan, Look Who s Talking: ECB Communiaion During he Firs Years of EMU Jaap Bos and Mindel van de Laar, Eplaining Foreign Dire Invesmen in Cenral and Easern Europe: an Eended Gravi Approah Jaap Bikker and Paul Mezemakers, Is bank apial prolial? A ross-ounr analsis Philipp Maier, EMU enlargemen, inflaion and adjusmen of radable goods pries: Wha o epe Karel-Jan Alsem, Seven Brakman, Le Hoogduin and Gerard Kuper, The Impa of Newspapers on Consumer Confidene: Does Spin Bias Eis?