CESfo Workng Paper Seres THE MIIMUM IFLATIO RATE FOR EUROLAD Hans-Werner Snn Mchael Reutter* Workng Paper o. 377 December 2000 CESfo Poschngerstr. 5 81679 Munch Germany Phone: +49 (89) 9224-1410/1425 Fax: +49 (89) 9224-1409 http://www.cesfo.de
CESfo Workng Paper o. 377 December 2000 THE MIIMUM IFLATIO RATE FOR EUROLAD Abstract As a result of the Balassa effect relatve prces change rapdly between and wthn the euro countres. Thus t s mpossble to fnd a common monetary polcy that wll result n prce stablty n all countres. Based on emprcal estmates of the Balassa model, the paper calculates a mnmum aggregate nflaton rate whch s compatble wth the requrement that no country face a deflaton. Ths mnmum aggeragate nflaton rate s 0.94% n the euro-11 countres and 1.13% n an extended Europe whch ncorporates the east European countres. Keywords: Inflaton target, Balassa-Samuelson effect, ECB JEL Classfcaton: E31, E52, E58 Hans-Werner Snn CESfo Poschngerstr. 5 81679 Munch Germany emal: SI@cesfo.de Mchael Reutter CESfo Schackstr. 4 80539 Munch Germany
1. Introducton What s the European Central Bank s (ECB) approprate nflaton target? Many observers argue that what used to be Europe s most stable currency, the deutschmark, should serve as the benchmark for the euro. The nflaton rate n Euroland should be as low as t was n Germany, because only then would the euro be able to replace the deutschmark and successfully compete wth the dollar. Ths paper contrasts ths vew by pontng to the dfferences between Euroland and Germany. Euroland s bgger and more dverse than Germany, and for the tme beng substantal relatve prce changes have to be accounted for. These relatve prce changes requre an nflaton rate for Euroland that s hgher than the one approprate for any sngle European country f the rsk of local deflaton s to be avoded. If, say, Ireland needs a prce ncrease relatve to Germany, then t makes lttle sense to prescrbe Germany s prevous dea of prce stablty to the average European prce ndex, because t mght requre deflaton for Germany or at least a dangerously low nflaton rate. Thus, the Bundesbank s goals should not be taken over by the ECB on a one-to-one bass. In ths paper we try to calculate the mnmum average nflaton rate for Euroland that prevents deflaton n any one of ts members. We do not clam that the ECB should target at ths mnmum nflaton rate; however, we argue that the mnmum nflaton rate s a specal component n the calculaton of the nflaton target that comes n addton to the usual arguments used n the closed economy context. It s a matter of fact that the sngle European countres have not, n the past, amed at prce stablty n the strct sense of the word, and they may have been wse n not dong so. There are at least four economc reasons why a sngle country should am at a postve measured nflaton rate. - There may be stuatons where a negatve real nterest rate s requred to get the economy gong (Summers 1991). However, when the nflaton rate s zero, negatve real nterest rates are mpossble, as negatve real nterest rates would then requre negatve nomnal nterest rates, but negatve nomnal nterest rates are excluded by the lqudty trap. The recent example of the Japanese recesson has demonstrated the valdty of ths concern. - In lght of nomnal prce rgdtes some nflaton s needed to allow for a declne of relatve prces and to provde the economy wth the necessary flexblty to accommodate structural change (Akerlof, Dckens and Perry 1996). Expermental studes (Fehr and Falk 1999) as well as ntervews (Kahnemann, Knetsch and Thaler 1986, Agell and Lundborg 1995, Bewley 1995, 1999) have demonstrated the mportance of such rgdtes. 2
- An nflaton tax on the holdng of real money balances should be part of an optmal tax system because ts dstortons balance wth the dstortons of other taxes (Phelps 1973) at the margn. - The measured nflaton rate may overestmate the true nflaton rate by between 0.5 and 2 percentage ponts, because the avalable statstcs count prce ncreases for qualty mprovements as nflaton. Ths effect has been frequently emphassed by Alan Greenspan (1998) and others (Boskn et al. 1996, Moulton 1996, Shapro and Wlcox 1996). All of these reasons for allowng for some degree of measured nflaton n each sngle European country wll reman vald n the future, but they are augmented by an addtonal reason that results from the need to allow for relatve prce changes between the European countres, and ths s what we analyse n ths paper. Our approach s based on the Balassa-Samuelson effect. 1 There are traded goods and non-traded goods, and there are advanced and less advanced countres. Traded goods have the same prce everywhere, but the prces of non-traded goods may dffer between the countres. Labour productvty n the traded goods sectors of the less advanced countres tends to grow exceptonally fast, drvng up wages n these countres. The wage ncrease translates nto an ncrease n the prce of non-traded goods relatve to the prce of traded goods and also to an ncrease n the average prce level of these countres relatve to the average prce level n the more advanced countres. Thus, a stable aggregate prce level for all countres taken together requres a deflaton n the more advanced countres, and a constancy of the more advanced countres prce level mples that there s nflaton n the aggregate. 2 The Balassa-Samuelson effect s an emprcal regularty rather than a theoretcal necessty. It was observed n less developed countres and explans why these countres have undervalued currences f seen from a comprehensve purchasng power party perspectve. Whle there are no less developed countres accordng to the usual defnton n Euroland, we fnd that the Balassa-Samuelson effect has nevertheless been very strong n prevous years. Some countres dsplay very large dfferences n productvty growth between the two sectors, others do not. We argue that these dfferences are lkely to result n a dvergence of the natonal nflaton rates of up to 2.7 percentage ponts and estmate a requred mnmum Euroland nflaton rate n the order of 1%. Inflaton n Euroland should therefore not be 1 Balassa (1964) and Samuelson (1964). 2 Although the nfluence of the Balassa-Samuelson effect on nflaton dfferentals n a monetary unon s acknowledged by the ECB, the ECB does not nfer any consequences for monetary polcy from ths fact (see European Central Bank, 1999). 3
compared wth the prevous nflaton n the deutschmark zone wthout allowng for such a mark-up that captures the prevalng dfferences n the European productvty growth rates. 2. Some Basc Arthmetc In ths secton we set up a two sector model for the area of a currency unon. Assume there are = 1,..., n countres whch are members of a currency unon. All countres have two sectors, one produces a tradable manufacturng good, the other produces a non-tradable good. Tradable goods have the same prces n all countres: 3 T T p = p = 1,..., n. (1) The prces of non-tradables, p, by contrast, can dffer between the countres. Let T y and y ndcate the margnal products of labour n the tradng and non-tradng sectors of country. Internal competton and free mgraton of labour between the two sectors mply that y p = w = y p, (2) T T where w s the common wage rate of country. Let a hat ndcate growth rates. Then t follows from (1) and (2) that the prce of country s non-traded goods grows at the rate pˆ ˆ ˆ ˆ T T = p + y y ; (3).e., t s gven by the rate of ncrease n the common nternatonal prce of the traded goods plus the dfference between country s productvty growth rates of the traded and non-traded sectors. Equaton (3) reveals already the basc mechansm explored n ths paper. If the ntersectoral growth dfference s hgher n some countres than n others, the prces of nontraded goods wll have to grow faster n these countres, and accordngly the nflaton rates of these countres wll have to be hgher than that n other countres. If deflaton n any one of the countres s to be avoded, the aggregate nflaton rate for all countres taken together wll 3 All varables depend on tme, t, but the tme ndex s omtted for the sake of brevty. Unless otherwse stated, all varables of an equaton hold for any arbtrarly gven pont n tme. 4
have to be hgher the more dverse the countres are n terms of ther ntersectoral growth dfferences as measured by yˆ yˆ. T To explore ths further, let us smplfy by assumng that aggregate prce ndces can be wrtten as geometrc means of the respectve sub-ndces. Then the nflaton rate of country, π, s gven by π ˆ pˆ (4) T ( 1 α ) p + α where α s the share n value added produced by country s sector of non-traded goods, and 1 α s the share n value added produced by ts sector of traded goods. Smlarly, the aggregate nflaton rate of all countres, π, s gven by n n π Σ β π, β =1 Σ =1 = 1, (5) where β s country s share n aggregate value added. Insertng (3) nto (4) we obtan T T π = pˆ +α ( yˆ yˆ ). (6) Equaton (6) reveals that country s nflaton rate equals the rate of ncrease n the prce of traded goods plus the share n value added of the sector of non-traded goods tmes the excess productvty growth of the sector of traded goods. If none of the n countres s allowed to experence a deflaton, π 0 =1,..., n, (7) t s obvously necessary that pˆ T T α ( yˆ yˆ ) =1,..., n. (8) Ths nequalty gves a lower bound for the rate of ncrease n the prce of non-traded goods. If productvty growth n the sector of traded goods exceeds that n the sector of non-traded goods n all countres, the rght-hand sde of nequalty (8) s negatve ndcatng that a 5
reducton of the common prce of traded goods s compatble wth the condton that there be no aggregate deflaton n any of the countres. If, on the other hand, there happens to be a country where productvty growth s hgher n the sector of non-traded goods than n the sector of traded goods, the rght-hand sde of the nequalty s postve and t follows that the prce of traded goods has to ncrease f a deflaton s to be avoded. Ths has strong mplcatons for the aggregate nflaton rate of all countres. From (5) and (6) t follows that the aggregate nflaton rate s gven by π = pˆ T n + Σβ α = 1 ( yˆ T yˆ ). (9) Let πmn denote the mnmum aggregate nflaton rate compatble wth the requrement that no sngle country wll face a deflaton. It s easy to show that (9) and (8) mply π mn n = Σ β α = 1 ( yˆ T yˆ ) mn = 1,..., n T [ α ( yˆ yˆ )]. (10) Ths equaton wll serve as the bass of our emprcal estmates. It gves precse meanng to the ntuton reported n the ntroducton. To nterpret the equaton, suppose frst that there s one mature country whose manufacturng sector enjoys no partcular productvty gans relatve to the sector of nontraded goods and whch therefore experences no change n relatve prces, whle the relatve prce of non-tradables s rsng n all other countres. In ths case, the prce of traded goods cannot fall wthout generatng a deflaton n the mature country. It follows that the aggregate prce level must rse so as the accommodate for the requred ncreases n the relatve prces of non-traded goods n the other countres that result from the productvty ncreases n the traded-good sectors; ths s reflected by the frst term on the rght-hand sde of the equaton. Suppose next that even the most mature country experences some extra productvty growth n the sector of traded goods. In ths case the prce of traded goods can fall wthout mplyng a deflaton n ths country because t s compensated for wth an ncrease n the prced of non-traded goods. The no-deflaton constrant s relaxed, and the relatve prce of non-traded goods can ncrease n all countres wthout mplyng as much nflaton as before, because the prce of traded goods s fallng. 6
3. Emprcal Results for Euroland In 1999, the frst year of the currency unon, the nflaton dfferentals between countres n the euro area were qute substantal. In Fgure 1 we plot the nflaton rates of the 10 major countres aganst the output gap, accordng to OECD defnton. As can be seen, the state of the busness cycle delvers only a partal explanaton for the dfferentals. Large resduals reman whch pont at the mportance of other nfluences such as the Balassa-Samuelson effect dscussed above. If the Balassa-Samuelson effect s robust, a postve mnmum nflaton rate on average s obvously necessary to prevent deflaton n any one country. Fgure 1: Inflaton Rates and Output Gap n 1999 3 Inflaton Rate 1999 2.5 2 1.5 1 0.5 Italy Ireland Span Portugal etherlands Fnland Belgum France Germany Austra 0-5 -4-3 -2-1 0 1 2 3 4 Output Gap 1999 Source: Monthly Reports of the German Bundesbank, February 2000 and OECD Statstcal Compendum, Economc Outlook. Legend: Inflaton rates are annual HICP nflaton n percent. Output gap accordng to OECD methods. Fndng the mnmum aggregate nflaton rate s dffcult snce, naturally, the avalable data stem from a tme where nether a common currency nor a fxed exchange rate prevaled and snce t s not clear how traded and non-traded goods should be dstngushed emprcally. Even the most local good s tradable when toursm and labour mgraton are taken nto account. Our decson of how to separate the sectors was drven manly by data avalablty. To construct a complete sample for all members of the currency unon, we had to restrct the sector of tradable goods to agrculture and manufacturng and to defne the remander as the 7
sector of non-tradable goods. on-tradables accordng to our defnton nclude constructon, servces, electrcty, gas and water supply. 4 We consder two sample perods, 1987 1995 and 1978 1996. In the frst perod we have data from all countres except Luxembourg, whch s of neglgble sze. We hope that ths perod s suffcently long to elmnate the role of busness cycles. To check the valdty of ths hope we repeat our calculatons wth the second, much longer perod for the subset of those sx countres for whch data were avalable. Table 1: Growth Rates of Labour Productvtes and Value Added Prces Country Labour Productvty Value Added Prces Dfference (1) (2) (3) (4) (5) (6) Traded on-traded Traded on-traded Productvty* Prces** Austra 3.21 (2.70) 1.07 (0.90) 1.38 (1.95) 3.41 (4.17) 2.14 (1.80) 2.03 (2.22) Belgum 3.07 (3.99) 1.74 (1.35) 1.62 (2.16) 3.35 (4.40) 1.33 (2.64) 1.73 (2.24) Germany 1.90 (1.72) 1.55 (1.27) 1.71 (2.51) 3.25 (3.49) 0.34 (0.46) 1.54 (0.98) Span 1.92-0.36 3.49 5.90 2.28 2.41 Fnland 5.98 (4.95) 1.88 (1.73) 1.62 (3.74) 4.45 (6.46) 4.10 (3.22) 2.84 (2.72) France 3.01 (2.98) 0.97 (1.09) 1.20 (4.54) 2.98 (6.01) 2.04 (1.89) 1.77 (1.47) Italy 3.79 (4.26) 1.50 (1.13) 3.34 (7.17) 5.79 (9.87) 2.29 (3.14) 2.44 (2.70) etherlands 2.91 0.79 0.89 1.90 2.13 1.02 Portugal 3.52 2.00 8.25 10.43 1.52 2.18 Ireland 6.07 1.84 0.23 3.76 4.23 3.52 * Traded mnus non-traded ** on-traded mnus traded otes: The numbers are arthmetc means of the annual growth rates for the perod 1987 1995 and, n parentheses, for the perod 1978 1996. Data for Germany refer to West Germany untl 1993. The sector of traded goods comprses agrculture and manufacturng, the sector of non-traded goods the remander. Labour productvty s defned as value added dvded by employment. Employment, nomnal value added and real value added are taken from the OECD Internatonal Sectoral Database and the OECD atonal Accounts. The excepton s Ireland for whch data on nomnal value added and employment were suppled by the Federal Statstcal Offce of Germany. Growth n Irsh real value added was proxed for the manufacturng sector by usng the ndex of ndustral producton from the OECD Economc Outlook. Irsh agrculture data were taken from the OECD Economc Accounts on Agrculture. In a frst step, we calculate the growth rates of labour productvty, equatng the growth of margnal and average labour productvtes. The results of ths calculaton are reported n table 1. As expected we fnd that productvtes grow faster n the sectors of traded goods than n the respectve natonal sectors of non-traded goods. Ths holds true for all countres n our sample. A formal statstcal test frmly rejects (p-value < 0.01) the null hypotheses that, across the countres, the productvty growth rates of the two sectors are dentcal. 4 Ths s n lne wth the usual practce n the emprcal lterature that tests the Balassa-Samuelson hypothess. See Canzoner, Cumby and Dba (1999). 8
Table 1 reveals a rather large nternatonal varance n the ntersectoral productvty growth dfferentals (column 5), ndcatng a potentally mportant role of the Balassa-Samuelson effect. The two extremes are marked by Germany and Ireland. Whle Germany experenced a productvty lead of the sector of traded goods over the sector of non-traded goods of only 0.34%, the lead was a remarkable 4.23% n Ireland over the sample perod. Gven that Germany had a much hgher labour productvty than Ireland and that Ireland has become a full member of the EU, t s plausble to observe a rapd catchng-up process. As the numbers n parentheses show, a longer sample does not produce substantally dfferent results. The choce of sample length mples a trade-off between smoothng short run fluctuatons (long sample) and keepng track of persstent changes (short sample). We use the short sample results for our subsequent calculatons. The growth rates of the sector-specfc prces gven n the same table mrror the productvty fgures. As expected, n all countres the prces of non-traded goods grew faster than the prces of traded goods. Accordngly, the hypothess that the cross-country averages of the rates of prce ncrease of tradables and non-tradables are dentcal s rejected at the fve percent level. When dfferent countres are compared, there s a rather close correlaton between the ntersectoral dfferences n productvty growth rates and the ntersectoral dfferences n prce changes, especally for the long sample. Ths s documented n columns 5 and 6 of table 1. We conclude that the Balassa-Samuelson effect s ndeed an mportant drvng force behnd the observed relatve prce changes n Europe. 5 We therefore proceed wth our model and derve the mnmum nflaton rates gven n equatons (6) and (10). The results of the calculatons are reported n the frst column of table 3 (see further below). The column reports the respectve natonal mnmum nflaton rates n the euro area and ther GDP-weghted averages. Ths average, 0.94%, s the mnmum nflaton rate of the euro-11 countres whch we sought. The crtcal country s Germany. Germany has an ntersectoral productvty growth dfferental of only 0.34%, whch s substantally below the dfferental of 4.23% n Ireland. To prevent Germany from experencng a general deflaton, the prce level of manufacturng goods throughout the euro area must not fall by more than 0.26% per year. Gven the hgh rates of prce ncrease for non-tradables n the countres catchng up, n partcular Fnland, Ireland, Italy and Span, ths mples an aggregate mnmum nflaton rate for all sectors and countres of roughly one percent. Ths aggregate mnmum nflaton rate s an average of very dfferent natonal rates. When Germany s GDP deflator s taken to reman constant n the 5 Ths s also confrmed by econometrc studes (see Canzoner, Cumby and Dba, 1999). 9
calculatons, Fnland needs an nflaton rate of 2.74%, and Ireland needs one of 2.35%. Ths shows how meanngless the attempt to stablse the prce level n all countres would be. Comparng the nflaton predctons based on our productvty comparsons wth the actual nflaton rates n 1999 (see Fgure 1), we dentfy two outlers. One s Portugal, whch has a much hgher nflaton rate than we should expect from ts rather low productvty dfferental. The other s Fnland, whch s far below the top poston n the nflaton rank that our productvty results predct. The explanaton of the hgh Portuguese nflaton rate can be sought n the extraordnary expanson of aggregate demand resultng from generous EU subsdes. Portugal takes the second rank after Israel among all OECD countres wth regard to the surplus of mports over exports at 13% of GDP (Snn and Westermann, 2000). Fnland, on the other hand, was ht by a very serous recesson n the early 1990s. One of the effects of ths recesson was a strong ncrease n unemployment and excess capactes, whch made a subsequent growth wth very low nflaton rates possble (Honkapohja and Koskela 1999). The other countres, however, have nflaton rates that are roughly n lne wth our predctons. Clearly, we cannot expect that our predctons are very precse for any partcular year, but the broad rankng of countres and, more mportantly, the dsperson n nflaton rates s roughly n lne wth our model. If nflaton rate dfferentals are below productvty dfferentals, the reason may, among other thngs, be found n nadequate monetary polces of the (see also Secton 5, Conclusons). In general, there s a noteworthy dfference n the nflaton rates predcted for the euro countres. Seven out of eleven countres wll need an nflaton rate whch s more than 1.3 percentage ponts above the rate approprate for Germany. We do not know what the latter really s. Germany s average nflaton rate over the perod 1985-1998 was 2.5%. If we assume that ths rate wll contnue to be approprate for Germany, then the seven countres wll need an nflaton rate of at least 3.8%, and the average Euroland nflaton rate wll have to be about 3.5%. If, on the other hand, we accept the Bundesbank s long standng goal of keepng the nflaton rate at 1.5%, the average Euroland nflaton rate wll have to be 2.5%. Both fgures are above the 2% target announced by the ECB and mght therefore lead to a reconsderaton of ths target. It s clear from the Balassa-Samuelson model that t would be false to mpose the ECB target on each sngle member country. Accordng to our calculatons, fast growng countres lke Span, Ireland or Fnland have mnmum nflaton rates of 1.5%, 2.4% and 2.7%. If the ECB adopted a GDP nflaton target of 1.5% for Germany t would have to tolerate that the nflaton rates of these countres approach 3.0%, 3.9% and 4.2%, respectvely. Such rates 10
would not be a sgn of an unsound economc polcy but could well be the natural mplcaton of the Balassa-Samuelson effect Our calculatons are rather senstve wth regard to the lowest ntersectoral dfference n productvty growth rates that one fnds among the countres consdered. The lower the lowest country s ntersectoral dfference n productvty growth, the lower s the relatve prce ncrease of non-traded goods n ths country and the lower the common nternatonal deflaton rate of traded goods whch s compatble wth prce stablty n ths country. As mentoned, the crtcal country s Germany n ths regard where the dfference n productvty growth s only 0.34%, accordng to our estmate. Germany s the country wth the hghest per capta GDP among the countres consdered. Thus t s plausble that t has the lowest rate of productvty growth n the manufacturng sector and the lowest lead n productvty growth over the sector of non-tradables. Over the sample perod, the German nflaton dfferental s larger than the productvty dfferental. Ths fact could be taken to mply that the relatve prce effect predcted by the Balassa-Samuelson model leads to an overstatement of the mnmum nflaton rate for Europe. ote, however, that the sample perod covers the tme after German unfcaton where the huge artfcal wage ncrease n east Germany mpled a rapd prce ncrease of non-traded goods n east Germany whch wll not be repeated n the future. Our productvty based estmates are free from ths dstorton and may therefore better predct the mnmum nflaton rate than a calculaton based on prces would have done. It s confrmng to see that before unfcaton, durng the perod 1978-1990, the average ntersectoral nflaton dfferental happened to be exactly equal to the 0.34% productvty dfferental on whch our calculatons are based. If we used ths number for the German ntersectoral nflaton dfferental and added the other countres actual nflaton dfferentals over the whole perod, as reported n table 1, a mnmum nflaton rate of 0.87% would emerge. Ths s rather close to our Balassa-Samuelson estmate of 0.94% and strengthens the confdence n our result. 4. Expandng the Monetary Unon Euroland may expand n the future. The frst new country to jon wll be Greece. The UK s lkely to have a general poll on membershp n the year 2002 after the next electon, and f the UK jons t s possble that Denmark and Sweden wll reconsder. The next step could be the enlargement of Euroland towards eastern Europe. By 2004, the frst fve east European countres may have joned the EU, and shortly thereafter another fve eastern countres may 11
become EU members. Also, Turkey, Cyprus and Malta are watng n lne. Of course, membershp n Euroland wll not automatcally follow EU membershp, but t s lkely to do so n the foreseeable future. Thus t may be useful to extend our analyss by ncorporatng alternatve extenson scenaros. We frst consder the ncluson of Greece, the UK, Denmark and Sweden. Table 2 shows the ntersectoral dfferences n productvty growth n these four countres. Obvously, the Brtsh and Dansh values are qute smlar to those of the other member countres, but Greece and Sweden have rather hgh values that even exceed those of Ireland. Table 2: EU Enlargement Country Labour Productvty Dfference Traded ontraded Denmark 4.10 2.00 2.09 UK 3.07 1.81 1.26 Greece 5.29-0.84 6.13 Sweden 6.56 1.28 5.28 Turkey 3.34 2.78 0.57 Poland 10.27 3.89 6.38 Hungary 11.07 0.82 10.24 Czech Republc 6.44 1.58 4.86 Estona 7.67 1.89 5.79 Slovena 7.28 1.80 5.48 otes: The numbers are arthmetc means of the annual growth rates. The traded sector comprses agrculture and manufacturng. The sector of non-traded goods s the remander. Labour productvty s defned as value added dvded by employment. The sample s from 1991-1997 for Denmark, UK, Sweden and Turkey, from 1991-1996 for Greece, from 1994-1997 for Poland and Hungary, from 1995-1998 for the Czech Republc, from 1994-1998 for Estona, and from 1996-1999 for Slovena. Data sources are the OECD Economc Outlook, World Bank Development Indcators, OECD atonal Accounts, OECD Annual Labour Force Statstcs, the Federal Statstcal Offce n Germany, IMF country reports and the Slovenan Economc Mrror. The expanson of EMU membershp leads to a further ncrease n the requred mnmum nflaton rate for Europe as s shown n table 3, second column. The average euro-15 nflaton rate that would prevent a deflaton n Germany s now 1.05%. The countres wth the hghest mnmum nflaton rates would be Sweden and Greece wth 3.64% and 4.26% respectvely. The UK would experence rather low mnmum nflaton, below the EU average, whle Denmark would be n the mddle feld of all countres. In a second step t can be envsaged that a number of further countres from eastern and south-eastern Europe become members of the currency unon. We therefore extended the calculatons to Poland, Hungary, the Czech Republc, Estona, Slovena and Turkey, omttng 12
Cyprus, Bulgara, Romana, Lthuana, Latva, Malta, and the Slovak Republc due to data problems. Labour productvty growth n these countres s also gven n table 2. As expected we observe rather hgh ntersectoral growth dfferences n all of the transton states, but agan the pattern s not homogeneous across the countres. Whle Turkey surprsngly has an ntersectoral growth dfference lke the western European countres, Hungary has one above 10%. Table 3: Countres Ranked by Mnmum Inflaton Rates Compatble wth the Absence of Deflaton n any one Country EU 11 EU 15 EU 21 Euro area 0.94 Euro area 1.05 Euro area 1.13 Germany 0.00 Germany 0.00 Germany 0.00 Belgum 0.80 UK 0.71 Turkey 0.09 Portugal 0.82 Belgum 0.80 UK 0.71 France 1.34 Portugal 0.82 Belgum 0.80 Austra 1.42 France 1.34 Portugal 0.82 etherlands 1.43 Denmark 1.39 France 1.34 Italy 1.49 Austra 1.42 Denmark 1.39 Span 1.53 etherlands 1.43 Austra 1.42 Ireland 2.35 Italy 1.49 etherlands 1.43 Fnland 2.74 Span 1.53 Italy 1.49 Ireland 2.35 Span 1.53 Fnland 2.74 Ireland 2.35 Sweden 3.64 Fnland 2.74 Greece 4.26 Czech Rep. 2.88 Slovena 3.38 Sweden 3.64 Estona 4.06 Poland 4.16 Greece 4.26 Hungary 6.86 otes: Mnmum nflaton rates mpled by by equatons (6) and (10) based on the estmatons of productvty dfferences between traded and non-traded sectors reported n tables 1 and 2. The table reports three dfferent estmates for dfferent sets of countres. All estmatons exclude Luxembourg. The weghts of the sectors and countres were based on value added shares n 1995. As s shown n table 3, column 3, the ncluson of these countres n the monetary unon would produce another ncrease n the mnmum aggregate nflaton rate from 1.05 to 1.13 %. Ths s not a dramatc change n lght of the hgh predcted nflaton rates n the order of 3-4% for most of the transton countres. The bgger the unon, the smaller the weght that a sngle country has. If the transton economes do not grow at a rate substantally above the EU average, they wll not have an mportant mpact on the average nflaton rate n Euroland n the foreseeable future. 13
5. Conclusons In ths artcle we have argued that the prevous nflaton targets of the Bundesbank should not be adopted by the European Central Bank because Europe s dversty n natonal productvty growth rates mples substantal relatve prce changes among the dfferent countres. To allow for these changes wthout causng any one country to deflate, the common monetary polcy has to be looser than the prevous German one. We fnd that the ECB has to tolerate roughly one addtonal percentage pont of nflaton n the aggregate beyond what had been consdered approprate for Germany alone. The Bundesbank tradtonally had amed at an nflaton rate of 1.5% to accommodate for Germany s nternal structural change. If ths target remans vald for Germany wthn the European currency unon, the euro-11 countres need an aggregate nflaton target of about 2.5% or more. Ths has sgnfcant mplcatons for some of the fast growng economes lke Span, Ireland or Fnland. If these countres nflaton rates le far outsde the range whch the Bundesbank would have tolerated n Germany, there s no reason to complan. Rates of 3%, 3.9% and 4.2%, respectvely, would be natural mplcatons of the Balassa-Samuelson effect, ndcatng, n fact, natural changes n relatve prces rather than a dangerous weakenng of the euro. The stuaton wll not change sgnfcantly f, at some stage, the new EU members n eastern Europe also jon the monetary unon. The mnmum European nflaton rate necessary to prevent deflaton n any one country wll ncrease slghtly but reman n the neghbourhood of 1%. However, countres lke Sweden, Poland or Hungary wll then need nflaton rates of 3.6%, 4.2% and 6.9%, or f the Bundesbank s 1.5% rule remans vald for Germany, even rates of between 5% and 8%. Our predctons of Europe s future changes n relatve prces are based on the Balassa- Samuelson model. Other models for a predcton of relatve prce changes are concevable, ncludng smple extrapolatons of prevous prce changes or sophstcated general equlbrum models of the European economy. We do not want to argue that the Balassa-Samuelson model s the only reasonable approach to the problem we consdered. However, we beleve that there s a systematc dfference among the European countres whch can better be detected by lookng at the productvty fgures rather than the prce changes drectly. Smply extrapolatng prevous prce changes runs the rsk of ncorporatng the effects of nadequate monetary polces n the past. If European monetary polces were too tght under Germany s monetary leadershp, as some authors ncludng Solow (2000) have argued, and f some nflaton s 14
necessary to allow for relatve prce changes to occur, as the emprcal approaches cted n the ntroducton have shown, then t makes lttle sense to use the relatve prce changes of the past to derve approprate nflaton targets for the future. Such an approach would run the rsk of severely underestmatng the necessary changes n relatve prces and hence the mnmum nflaton rate for Europe. In the ntroducton we summarsed several arguments for why very low nflaton rates mght be harmful for an economy. If the ECB does not ncrease ts current target of two percentage ponts, Germany may have to lve wth a measured natonal nflaton rate of only one percent. If the measurement error n connecton wth overlooked qualty mprovements s taken nto account, ths may mply that the actual German consumer prce ndex wll stay constant or even shrnk. Ths would be a dangerous stuaton, because Germany s rgd labour markets are especally prone to suffer from downward nomnal wage rgdtes ponted out by Akerlof, Dckens and Perry (1996). Cyclcal downturns mght turn out to be even more harmful to employment than they were n the past. Of course, the damage to Germany should be weghed aganst the possble advantages of havng low nflaton n Ireland or Fnland. However, as long as the evdence of economc costs of moderate nflaton rates remans as weak as t s, we conclude from our analyss that the ECB should tolerate a substantally hgher nflaton rate n Europe than the Bundesbank used to n Germany. 15
References Agell, Jonas, and Per Lundborg (1995), Theores of Pay and Unemployment: Survey Evdence from Swedsh Manufacturng Frms, Scandnavan Journal of Economcs 97, pp. 295-307. Akerlof, George A., Wllam T. Dckens, and George L. Perry (1996), The Macroeconomcs of Low Inflaton, Brookngs Papers on Economc Actvty 1, pp.1-76. Balassa, Bela (1964), The Purchasng Power Party Doctrne: A Reapprasal, Journal of Poltcal Economy 72, pp. 584-96. Bewley, Truman F. (1995), A Depressed Labor Market as Explaned by Partcpants, Amercan Economc Revew 85, pp. 250-54. Bewley, Truman F. (1999), Why Wages Don't Fall Durng a Recesson, Cambrdge, Mass.: Harvard Unversty Press. Boskn, Mchael J., Ellen R. Dulberger, Robert J. Gordon, Zv Grlches, and Dale W. Jorgenson (1996), Toward a More Accurate Measure of the Cost of Lvng (The Boskn Commsson Report), Fnal Report to the Senate Fnance Commttee. Canzoner, Matthew B., Robert E. Cumby, and Behzad Dba (1999), Relatve Labor Productvty and the Real Exchange Rate n the Long Run: Evdence for a Panel of OECD Countres, Journal of Internatonal Economcs 47, pp. 245-66. European Central Bank (1999), Inflaton Dfferentals n a Monetary Unon, Monthly Bulletn October 1999, pp. 35-44. Fehr, Ernst, and Armn Falk (1999), Wage Rgdtes n a Compettve Incomplete Contract Market: An Expermental Investgaton, Journal of Poltcal Economy 107, pp. 106-34. Greenspan, Alan (1998), Remarks at the Annual Meetng of the Amercan Economc Assocaton and the Amercan Fnance Assocaton, Chcago, Illnos, January 3, 1998. Avalable at: www.federalservce.gov/boarddocs/speeches/19980103.htm. Honkapohja, Seppo, and Erkk Koskela (1999), Fnland's Depresson: A Tale of Bad Luck and Bad Polces, Economc Polcy 29, pp. 399-436. Kahnemann, Danel, Jack L. Knetsch, and Rchard Thaler (1986), Farness as a Constrant on Proft Seekng: Enttlements n the Market, Amercan Economc Revew 76, pp. 728-41. Moulton, Brent R. (1996), Bas n the Consumer Prce Index: What s the Evdence?, Journal of Economc Perspectves 10, pp. 159-77. Phelps, Edmund (1973), Inflaton n the Theory of Publc Fnance, Swedsh Journal of Economcs 75, pp. 67-82. 16
Samuelson, Paul A. (1964), Theoretcal otes on Trade Problems, Revew of Economcs and Statstcs 23, pp. 1-60. Snn, Hans-Werner, and Frank Westermann (2000), Two Mezzogornos, CESfo Workng Paper o. 388. Shapro, Matthew D., and Davd W. Wlcox (1996), Msmeasurement n the Consumer Prce Index: An Evaluaton, n: Ben S. Bernanke and Julo Rotemberg (eds), BER Macroeconomcs Annual, Cambrdge, Mass.: MIT Press, pp. 93-164. Solow, Robert (2000), Unemployment n the Unted States and n Europe: A Contrast and the Reasons, CESfo Workng Paper o. 231. Summers, Larry (1991), How Should Long-term Monetary Polcy be Determned?, Journal of Money, Credt and Bankng 23, pp. 625-31. 17