NBER WORKING PAPER SERIES TRADE INVOICING IN THE ACCESSION COUNTRIES: ARE THEY SUITED TO THE EURO? Linda Goldberg

Transcription

1 NBER WORKING PAPER SERIES TRADE INVOICING IN THE ACCESSION COUNTRIES: ARE THEY SUITED TO THE EURO? Lina Golberg Working Paper NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambrige, MA September 2005 Feeral Reserve Bank of New York an NBER. Prepare for the International Seminar on Macroeconomics, National Bureau of Economic Research, Buapest June Roc Armenter, Charles Engel, Rebecca Hellerstein, Richar Portes, Ceric Tille, an participants at the ISOM meeting provie valuable comments. Eleanor Dillon provie excellent research support. The views expresse in this paper are those of the iniviual authors an o not necessarily reflect the position of the Feeral Reserve Bank of New York or the Feeral Reserve System. Aress corresponences to Lina S. Golberg, Feeral Reserve Bank of NY, Research Department, 33 Liberty St, New York, N.Y Lina.Golberg@ny.frb.org. The views expresse herein are those of the author(s) an o not necessarily reflect the views of the National Bureau of Economic Research by Lina Golberg. All rights reserve. Short sections of text, not to excee two paragraphs, may be quote without explicit permission provie that full creit, incluing notice, is given to the source.

2 Trae Invoicing in the Accession Countries: Are They Suite to the Euro? Lina Golberg NBER Working Paper No September 2005, Revise March 2006 JEL No. F3, F4 ABSTRACT The accession countries to the euro area are increasingly bining their economic activity, external an internal, to the euro area countries. One aspect of this phenomenon concerns the currency invoicing of international trae transactions, where accession countries have reuce their use of the US ollar in invoicing international trae transactions. Theory preicts that the optimal invoicing choices for accession countries epen on the composition of goos in exports an imports an on the macroeconomic fluctuations of trae partners, both bearing on the role of hering an heging consierations within exporter profitability. These consierations yiel country-specific estimates about the egree of euro-enominate invoicing of exports. I fin that the exporters of some accession countries, even in their trae transactions with the euro zone an other European Union countries, might be pricing too much of their trae in euros rather than in ollars, thus taking on excessive risk in international markets. Lina S. Golberg Research Department, 3r Floor Feeral Reserve Bank-New York 33 Liberty Street New York, NY an NBER lina.golberg@ny.frb.org

3 I. Introuction The accession countries to the euro area are increasingly bining their economic activity, external an internal, to the euro area countries. One aspect of this phenomenon concerns the currency invoicing of international trae transactions. There has been a substantial shift away from the use of the US ollar in accession country international trae transactions. In this paper, I explore the theoretical rivers of optimal invoicing choices for exporters, highlighting the importance of the composition of goos in exports an imports, an the partner composition of these forms of trae. I explore whether accession country exporters, by invoicing in euros an thus closely aligning their trae with that of the rest of the euro area, are pursuing economically appropriate strategies. Perhaps some of the accession country export transactions are not as well suite to euro invoicing, leaing proucers with overly high euro shares in pricing to expose themselves to excessive risk in international markets. The analysis raws on lessons from the theoretical moel of Golberg an Tille (2005), which presents the eterminants of the relative importance of heging motives an hering motives in currency invoicing (an exchange rate pass through) choices by exporters. This moel motivates an empirical application to the accession countries. The moel shows the role of macroeconomic volatility an inustry composition in exporter pricing strategies, emonstrating that optimal currency invoicing strategies consist of a mix of heging consierations an hering. Macroeconomic volatility consierations have been emphasize in a range of papers, from Giovannini (1988) through recent contributions by Devereux, Engel an Storegaar (2004), Oi, Otani, an Shirota (2004), an Engel (2005). By introucing an explicit role for elasticities of substitution in eman an ecreasing returns to scale in pricing, Bacchetta an van Wincoop (2005) an Golberg an Tille (2005) [GT] show that macroeconomic volatility may mainly play a role in pricing an currency invoicing ecisions for ifferentiate proucts. 1 Instea, they stress that inustry structure, emphasize early on by McKinnon (1979), is the key eterminant of how much hering occurs in pricing an currency invoicing ecisions. Proucers in highly competitive inustries, an proucers facing a high egree of ecreasing returns to scale in prouction, may optimally mimic the pricing strategies of their competitors in markets in which 1 While Bacchetta an van Wincoop iscuss this hering motive in the case where the exchange rate is the only source of volatility, Golberg an Tille also inclue volatility in wages an foreign eman. Campa an Golberg (2005) an Campa, Golberg, an Gonzalez-Minguez (2005) show that macroeconomic volatility is less important than inustry features in etermining exchange rate pass through into import prices. 1

4 their goos are sol. This leas to hering in invoice currency selection, while not explicitly seeking to ientify which currencies will be use in such hering. Various strategies have been offere elsewhere to pin own the equilibrium choice of hering currency. One example is through introucing network externalities in foreign exchange markets interacting with transaction costs in securities markets, as in Portes an Rey (1998). Using reasonable ranges of parameters, GT show that this hering activity coul be much more important in ecision making than the influence of heging an macroeconomic volatility. These consierations are applie to the trae transactions of eleven countries aspiring to join the euro area: Bulgaria, Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Polan, Slovakia, an Slovenia. A simple picture rawn from GT motivates the focus of the paper, which explores the extent to which accession country exporters use ollars, euros, or other currencies in invoicing their international trae transactions. Using mostly ata for 2002, GT compare the actual share of ollar use in invoicing country exports to what might be expecte purely on the basis of trae with the Unite States an the composition of country trae. The pattern of this relationship for a broa sample of countries is shown in Chart 1. Chart 1 works with the assumption that all exports to the Unite States an all exports of reference price an organize-exchange trae goos ( RW goos ) to other countries are invoice in ollars, the stanar currency for pricing most of these transactions. If this assumption were true, a country woul have its observations lie along the 45 egree line in this figure. Observe that many of the accession countries have invoicing patterns above the 45 egree line, suggesting that they have fewer exports invoice in ollars compare with what woul be preicte by this simple metric. Estonia, Slovenia, Slovakia, Hungary, an the Czech Republic all fit this escription. The exception is Latvia which, like Greece, ha consierably more use of ollars in invoicing exports than expecte purely on the basis of Latvia s trae with the Unite States an exports of highly substitutable goos. What are the consequences of exporters making invoice currency choices that are inappropriate, given observe inustry features an the volatility of macroeconomic conitions? These choices woul lea to lower expecte profits an more volatile profits than are optimal. As a hypothetical, consier the case of an accession country exporter invoicing in euros while competitors in euro area countries or in the Unite States are invoicing in ollars. Suppose as well that there is a high egree of prouct substitutability between the exporter s goos an its 2

5 competitors goos. With fluctuations in exchange rates, the exporter s relative price will vary ex post, even if ex ante the common currency expecte prices of the goos were ientical. Chart 1: Vehicle currency use of the ollar an commoity type exports Country's Exports to the US Share plus RW Goos Share of Total Country Exports % Bulgaria Cyprus UK Franc Estonia Spain Germany Luxembourg Slovakia Italy Polan Slovenia Czech Republic Hungary 45* Belgium Portugal Latvia US Dollar Invoicing Share in Country Exports, % Source: DOTS an various national sources Euro-area country ata, with the exception of Italy ata, refer to extra-euro area trae an invoicing Japan Australia Reprouce from Figure 5, Golberg an Tille (2005) ata use. Greece Malaysia Thailan Korea An unanticipate ollar appreciation against the euro woul lea the exporter to experience a sharp increase in eman for its goos. Given ecreasing returns to scale, the exporter woul also have an increase in his marginal costs. The net effect on expecte profitability will epen on the elasticities of substitution, the returns to scale in his prouction function, an a mix of covariances between the revenue an cost conitions he faces. In the reverse case of an unexpecte ollar epreciation against the euro, the exporter s goos are ex post excessively expensive, leaing to extensive substitution away from his proucts. This example shows that, even in cases where accession country exporters are overwhelmingly traing with euro area countries, ollars may still be more appropriate for invoicing trae with the euro area for those goos on which ollars are vehicle currencies for pricing. Inee, exporters in euro area countries use ollars for invoicing similar proucts even on trae transactions within the euro area, as recently iscusse by the European Central Bank (ECB, 2004). 3

6 Exploring these questions for the accession countries starts with the moel an lessons of the optimal invoicing currency work of Golberg an Tille (2005). After briefly presenting the theoretical motivation (section II), I present empirical etail on the irection of trae of accession countries with respect to euro area countries, the rest of the European Union, the Unite States markets, an other ollar bloc countries, followe by etail on the prouct composition of accession country exports an imports. For this work the Rauch (1999) inices, constructe to highlight the role of types of networks use in goos market transactions, are applie to etaile accession country export an import ata. This application prouces shares of goos in trae that are best escribe as ifferentiate, reference price, or price in an trae on organize exchanges. Organize exchange trae goos, such as commoities, are assume to be highly substitutable an the theory preicts that proucers of such goos shoul her in their invoicing choices. In practice, the ollar typically has been the currency use in such hering, at least in post war transactions. Conceptually, as long as this remains the case, the share of these types of goos in exports is treate as a lower boun on ollar invoicing. The paper s focus next turns to the secon consieration: the role of heging in optimal invoicing ecisions by exporters. The theory preicts that optimal heges shoul cover expecte shocks to proucer marginal costs. In other wors, for stable expecte profits, a proucer shoul invoice in a currency that yiels positive revenue shocks at the same time that the proucer faces high marginal costs, either because wages fluctuate or because aggregate eman fluctuates in an environment of non-constant returns to scale. An invoice currency shoul be selecte for heging purposes if it provies the highest covariance between an exchange rate (on the proucer export revenue) an marginal costs, or if it helps limit eman for a prouct at precisely the time when marginal costs of prouction are high. Thus, revenue will be high just at the times that marginal costs are high. For each country, we explore whether the ollar or the euro better suits this objective in export transactions to the Unite States, to the euro area, or to the rest of the European Union. A number of interesting observations are generate base on this analysis. First, there is consierable cross-country heterogeneity in the portion of exports invoice in ollars versus in euros. On average, more than half of the exports of accession countries are invoice in euros, while the average share of exports invoice in ollars is closer to 25 percent. Euro use has been increasing, with some of this euro gain matche by eclines in ollar use in export invoicing. 4

7 Inee, ollar use as an invoicing currency has ecline even for some countries that initially ha surprisingly low use of ollars by the rough analytical metrics of the paper. On accession country exports to countries other than the Unite States an the euro area, euro invoicing is roughly 37 percent, a lower share than observe for euro area countries. 2 The Unite States is not a major export estination for the goos of most accession countries, typically receiving less than 5 percent of these exports. Most of the accession country exports go to the euro area an the rest of Europe. Between 60 an 85 percent of the total exports of accession countries are characterize as ifferentiate proucts. While the remaining exports are often in reference price goos, for example paper, some countries also export substantial amounts of organize exchange trae goos like copper an aluminum. Much of the latter types of exports have ollar pricing worlwie. Controlling both for the structure of partners in trae an the composition of trae goos proucts, some accession countries use euros more heavily an use ollars less frequently in invoicing than o euro area countries. Examination of the optimal currencies in invoicing for heging purposes also yiels interesting conclusions for accession country exports to the Unite States, the euro area, the rest of the European Union, an Asia. The covariance analysis for this work compares the esirability of invoicing in euros versus ollars in exports to each estination market. In most cases, neither the ollar or the euro are appropriate choices for heging the eman an marginal cost risks to profits from exporter perspectives. Lithuania significantly favore the ollar as a heging currency prior to 2000, but this relationship then isappeare. Since 2000, neither the ollar or euro was theoretically a strong hege, except in the limite case of Hungarian exports to Asia, which woul have favore the euro. Mostly, base on heging, exporters in accession countries woul be inifferent to the euro or ollar as an invoice currency choice on their exports to Europe, the Unite States, an Asia. Overall, if the ollar is the key vehicle currency for many countries on goos we call reference price an exchange trae, our results suggest that some accession countries may have move further towar the euro in trae invoicing than is potentially optimal. This argument relies, in part, on an assumption that the ollar has retaine a central role as a vehicle currency in the goos that are reference price an trae on organize exchanges. The valiity of this 2 Small countries typically have low use of their own currencies in international trae transactions, as reporte in Golberg an Tille (2005). The two accession countries that report this information, the Czech Republic an Latvia, use their home currency on invoicing less than 10 percent of their imports an exports. 5

8 assumption, an its relation to exchange rate regimes, is iscusse in the concluing remarks of section IV. II. A three-country / three currency moel of invoice currencies The theoretical exposition closely follows the moel of Golberg an Tille (2005) [GT] on currency choice for trae invoicing. As exposite above, GT evelop the interaction between inustry features an macroeconomic variability in a new open-economy macro moel with three countries an price rigiities, builing on both Devereux, Engel an Storegaar [DES] (2004) an Bacchetta an van Wincoop (2005). While GT o not erive a general equilibrium version of the DES moel, they exten the existing theory of invoice currency selection in several critical imensions. First, GT move from the DES two-country / two-currency worl to a three-country / three-currency one, allowing for invoicing in a vehicle currency that belongs neither to the exporter nor the importer home market. Secon, GT evelop the contrasting roles in optimal invoice currency selection of inustry characteristics, such as the substitutability between competitors' goos, an macro-economic factors, such as business cycle an exchange rate volatility. The firm s incentive to limit the fluctuations of its relative price by choosing a trae invoicing strategy close to that of its competitors leas to a type of hering behavior in invoice currency choices for the exporters of relatively homogeneous proucts. This feature is also emphasize in Bacchetta an van Wincoop (2005). Thir, GT introuce ecreasing returns to scale in prouction, so that increases in output increase marginal costs even when wages are not responsive. Overall, GT conclue that macroeconomic variability is an important consieration in optimal invoicing only for trae in ifferentiate proucts. The egree of macroeconomic volatility neee to isturb an invoicing status quo for trae in more homogeneous proucts woul nee to be exceptionally large. The theoretical preiction is that -- even within a country where all economic agents face the same egree of macroeconomic volatility-- ifferent proucers will make ifferent invoice currency choices. Moreover, an exporter with two istinct traing partners is more likely to use istinct currencies on invoicing his exports to these istinct partners when his prouction is in ifferentiate goos an when he faces lower levels of ecreasing returns to scale in prouction. 6

9 The moel set-up. Before turning to the empirical implementation for accession countries, this section presents an abrige version of GT. An exporting firm is assume to have to post a price for its goos before knowing the realization of various shocks affecting the economy. The exporter is locate in country e, prouces a bran z, an sells her goos to the estination country. Goos are prouce using a technology with ecreasing returns to scale: (1) Y e 1 α ( z) = ( α ) [ H ( z) ], 0 < α 1 e where Y e ( z) is the output of z, ( z) The firm faces the following eman in estination country : ( z) [ P ( z) / P ] (2) Y = e C H e is the labor input, an α is the returns to scale parameter. -λ where C is the total eman for brans of the relevant sector in country, ( z) P e is the price, in country currency, of the bran z prouce in country e, an P is the price inex, in country currency, across all brans of the relevant sector sol in country. λ>1 is the elasticity of substitution between the various brans. Accoring to (2), the eman for a specific bran epens on its price, relative to the prices of other brans in the sector, an on the strength of overall eman in the estination market. The exporter proucing bran z sets its price in currency k, P k ( z) e, before the realization of the shocks affecting the economy. The currency of invoicing can be the currency of the country in which the exporter is locate (k=e), the currency of the country of estination (k=), a thir vehicle currency (k=v), or a combination of these three currencies. The exporter s price is set in currency k to maximize expecte profits represente by (3): k ( z ) = ED S P ( z) S ( z) ( α ) S k ek e ek e (3) Π α e e ek e C We C Se P Se P where P k λ S ek is the exchange rate between currency e an currency k, in terms of units of currency e per unit of currency k so that an increase correspons to a epreciation of currency e. state-specific iscount factor at which profits are evaluate, an 1 P k ( z) λ 1 α D e is the W e is the nominal wage. With 7

10 k its price set in currency k, the unit revenue for the exporter in currency e is S P ( z) -1 k the price in currency pai by consumers in the estination country is [ S ] S P ( z) e ek ek e e. Similarly,. Optimal invoice currency selection. Maximize profits are obtaine through the exporter choice of the currency k in which her goos are invoice. In making this selection, the exporter regars all the other variables in (3), such as the estination market eman, exporter wages, aggregate prices in the estination market an the bilateral exchange rate as exogenous to her invoicing ecision, with lower case variables enoting log eviations from the steay state x = ln X ln X ). Without constraining the exporter to invoice entirely in any currency e, or ( ss v, the invoicing ecision is a choice of weights of the three available currencies in the invoicing currency basket k. Specifically, the weights of currencies an v in the invoicing of exports to country are v β an β respectively, with the weight of currency e being 1 β β, an with the sum of the weights boune between 0 an 1. The case of pricing in one currency only is given by setting the weights to 0 or 1. Specifically, proucer currency pricing (PCP), which correspons to the proucer keeping unit revenues fixe in his own currency, correspons to β = v β = 0. Local currency pricing (LCP), in which the proucer has unit revenue stabilize in v the buyer s currency, correspons to β = 1, β = 0. Vehicle currency pricing (VCP) is given by v β = 0, β = 1. The sensitivity of p, the relative price between bran z an the competing brans,, to exchange rate movements plays a central role in the invoice currency choice. Some brans are invoice in currency, so the price pai by the consumers for these brans is unaffecte by exchange rate movements. Other brans are invoice in currency e, an the consumer price in currency moves with the exchange rate between the two currencies, s e v, with consumer paying a higher price when currency e appreciates (i.e s < 0). A final set of brans are invoice in currency v, so the price pai by consumers is higher when currency v appreciates (i.e. s e s ev < 0 ). We enote the total share of competing brans invoice in currency by e η, an e v the shares invoice in currency e an v by η an η respectively. In this case, the exporter s relative price of the goo sol in the estination market becomes: 8

11 k e v v ( β η ) se + ( β ) s ev (4) q = η Expression (4) shows that, while stabilization of unit revenues requires β =1, full stabilization of his relative price instea requires an exporter to choose weights on the ifferent currencies that exactly correspon to their shares in the inustry wie price inex: β = η, β = η. However, stabilization of the relative price is not the only consieration riving the exporter s ecision. Optimal invoicing weights v β an β maximize expecte profits uner the constraint v that β, β an β + β v o not fall outsie the [0,1] interval an given the structure of eman an costs shocks to which the exporter is subjecte. GT show that the optimal invoicing basket solution for the case where an exporter is selling only to one estination market is: 3 (5) (6) (7) where: The term β v e = Ωη + β = Ωη + v β = 1 β λ Ω = α + ( 1 Ω) ρ( me, se ) ( 1 Ω) ρ( me, sev ) v e β = ( 1 Ω) + Ωη ( 1 Ω) [ ρ( m, s ) + ρ( m, s )] ( 1 α ) 1 α, me = we + c λ( 1 α ) α m e entering into equations (5)-(7) is a covariance reflecting the influence of exogenous factors, for example exporter wages, prouctive inputs, an estination market aggregate eman, on the firm s marginal cost. Because of ecreasing returns to scale, a 1 percent increase in eman requires a e 1 / α percent increase in the labor input, hence a 1 / α percent increase in cost, holing the wage constant. The increase in eman also leas to a 1 percent increase in revenue, holing the price constant. The net increase in the marginal cost is then 1 1/ α = ( 1 α )/ 1 percent. The terms ρ ( m, ) an ( m, ) e s e coefficients that capture the covariances between marginal cost, an s ev. e ρ in (5)-(7) are regression e s ev m e e ev v v, an the exchange rates s e 3 Golberg an Tille (2005) erive a similar set of intuitions for the case where the exporter is constraine to use a single currency, rather than a basket of currencies, in his optimal selection. The results are qualitatively the same. 9

12 Components of optimal invoicing. While invoicing in the exporter s currency has the e avantage of fully stabilizing the exporter s marginal revenue, this full stabilization ( β = 1) is not necessarily an optimal choice for two reasons shown in equations (5)-(7). The first reason reflects a hering motive, capture by the terms Ω η an Ω v η. The exporter optimally limits the movements of her relative price by choosing an invoicing strategy close to that of her competitors: the exporter places a higher weight on invoicing in the estination currency, when her competitors have a higher share η of their own sales invoice in that currency. β, The secon motive for a proucer to move away from PCP is ue to heging, as capture by the terms ( 1 Ω) ρ( m, ) an( Ω) ρ( m, ) e s e 1. These terms measure the potential e s ev for an exporter to have an invoicing strategy that helps profits by limiting the impact of fluctuations in marginal costs on her profits. If she invoices in the estination currency,, a epreciation of her currency vis-à-vis the estination currency ( s > 0 ) increases unit revenue, in her own currency. If epreciations of this exchange rate ten to be correlate perios of increases in marginal costs, i.e. ρ ( m, ) > 0 e s e e, invoicing in the estination currency inuces a positive correlation between marginal revenue an marginal costs, reucing some of the volatility in profits. A similar logic applies to the vehicle currency. Inee, if we were to consier alternative vehicle currencies for use in an export transaction, the moel implies that the heging portion of the invoicing ecision shoul favor the currency (i.e. the bilateral exchange rate) that is significantly an most positively correlate with the shocks to exporter costs, regarless of whether these arise through prices of importe inputs, local currency wages, or fluctuations in aggregate estination market eman. The balance of influence on the hering imension versus the heging imension in (5)- (7) is given by the term Ω, which solely reflects the structural parameters of the moel, namely the elasticity of substitution between goos, λ, an the egree of returns to scale, α. The hering imension is more pronounce ( Ω is large) in inustries where goos are more substitutable (λ is large), since movements in relative prices then leas to large fluctuations in quantities sol. 10

13 The effect is also stronger when the technology exhibits larger ecreasing returns to scale (α is small), because fluctuations in output generate large movements in marginal cost. 4 Clearly, this theoretical exposition argues that optimal invoicing has both countryspecific an inustry-specific consierations. The country-specific macroeconomic correlations mainly apply to the exporters of highly ifferentiate proucts. By contrast, exporters in inustries proucing a more homogenous goo (i.e. goos that are more substitutable with those of their competition) woul optimize by following inustry practices an invoicing in a basket of currencies close to that of their competitors. 5 III. Invoicing Trae for Accession Countries A recent ECB report 6 provies ata on euro invoicing of imports an exports for eight euro zone countries (Belgium, France, Germany, Greece, Italy, Luxembourg, Portugal, an Spain), all ten newly accepte countries to the European Union (Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Polan, Slovakia, an Slovenia), an Bulgaria, a European Union caniate country. In the analysis below, Bulgaria is inclue with the euro area accession countries, misusing the terminology for brevity purposes. The ECB ata run from 2000 to 2003, with less complete coverage across countries in the early years. The ata from the ECB report are supplemente with ata on euro, ollar, an local country invoicing gathere from iniviual country sources, as etaile in the appenix tables of Golberg an Tille (2005). The accession country ata are presente in Table 1, with the top panel proviing broa etails for 2000, an the lower panel proviing etails for 2002, the last year for which ollar invoicing ata are wiely available. 4 If we were operating in an environment of increasing returns to scale, with α boune by negative 1 an zero, the role of λ woul be ampe as the egree of scale economies in prouction rises. The effect of an increase in returns to scale, i.e. an α that is more negative, is an unambiguous reuction in the invoicing weight on hering. 5 This theoretical exposition has focuse on invoice currency choice when prices are sticky over the invoicing interval. Of course, in some cases flexible prices may better match reality. In this case, our lessons still hol since there is a irect parallel between optimal invoice currency selection an observe levels of exchange rate pass through into trae goos prices [Golberg an Tille (2005) an Engel (2005)]. There is a irect corresponence between moels of optimal invoice currency selection uner sticky prices an those of partial exchange rate pass through in the case of flexible prices. 6 Review of the International Role of the Euro, Jan

14 Table 1: Dollar an Euro Shares of Trae Invoicing in Accession Countries Invoicing Patterns in 2000 Exports Imports Euro Share Dollar Share Euro Share Dollar Share average: Bulgaria Czech Republic* Latvia** Invoicing Patterns in 2002 Exports Imports Euro Share Dollar Share Euro Share Dollar Share average: Bulgaria Cyprus Czech Republic Estonia Hungary Latvia ** Lithuania Malta Polan Slovakia Slovenia * ata from 2001 instea of 2000 ata from 2003 instea of 2002 ** Latvian ata are for overall invoicing of imports an exports combine All shares are for invoicing of goos an services combine except for the Czech Republic (goos only). Source: ECB (2002, 2004, 2005) an iniviual country sources (etails in Appenix Table 2) Accession countries invoice their imports an exports largely in euros, with an average euro share well over 50 percent in However, the cross-country variation in the role of the euro in export invoicing is large, ranging from below 25 percent for Cyprus an Lithuania to over 60 percent for the Czech Republic, Estonia, Hungary, Polan, Slovakia, an Slovenia. With exception of Cyprus, the ata inicate a significantly smaller share of exports an imports invoice in U.S. ollars. Most accession countries o not report local currency invoicing shares. The two countries that o, the Czech Republic an Latvia, report home currency shares for imports an exports at or below 10 percent. Since the sum of euro an ollar are closer to 80 percent than 100 percent for some countries (Cyprus, Czech Republic, Estonia, Latvia, for 12

15 example), it is evient that currencies other than the ollar an euro still play a role in invoicing trae. Among these countries, only a few have invoicing ata publishe both for 2000 an 2002, thereby proviing only a limite perspective on how invoicing patterns are changing over time. The available ata are consistent with the euro growing in its role as the currency use in invoicing both export an import transactions. This pattern is shown in Chart 2, where the left most bars inicate the increase in average annual euro use in invoicing the exports of the accession countries for which we have 2000 an 2002 ata. To provie perspective on these evelopments, in the right-most bars I introuce comparable information for the euro area countries. 7 Among the three accession countries reporte, the biggest increase in euro share over 2000 through 2002 is for Bulgaria, at almost 8 percent annually, followe by Latvia at 6 percent, an the Czech Republic at uner 4 percent. The increase in euro use on export invoicing by euro area countries has been within a similar range over this time frame, again with variation across countries. Referring back to Chart 1, only Greece an Latvia ha an increase in euro use in invoicing that might be expecte base on the prior unexplaine large vehicle role of the ollar in its exports. 8 7 The ata for the accession countries an Italy cover all exports, while the ata for the other euro area countries cover extra-euro area exports only. 8 Appenix Table 1 provies the raw ata on invoicing for euro area countries. 13

16 Chart 2: Average Annual Rise in Euro Invoicing of Exports, * Change in Euro Invoicing Share (in % share of total exports) Bulgaria Czech Latvia Belgium France Greece Accession Euro-zone * or longest available perio. Details of time perios in Table 1. The ata cover all exports for accession countries an Italy; otherwise, extra euro area exports. Source: Author s calculations using ata from the ECB report an local country sources (etails in appenix) Italy Portugal Spain Chart 3 shows the extent to which increasing uses of euros in export invoicing came through reuction in the use of U.S. ollars for these purposes. For those countries for which relevant information is available, there has been both an (average annual) increase in the euro an an (average annual) ecline in the ollar in export invoicing. Among the three accession countries for which there is appropriate ata, euro gains have roughly matche ollar eclines for Bulgaria an Latvia. For the Czech Republic, where ollar share in invoicing starte low, increase euro use came through reuce use of other currencies. The experience among the euro area countries has also been mixe. For Greece, Spain an Italy, most of the increase in euro use came in parallel to reuce use of ollars. This was not the case for France, an was only partially the case for Portugal. 14

17 10.00 Chart 3: Average Annual Rise in Euro Invoicing an Fall in Dollar Invoicing of Exports, Change in Invoicing Share (in % of total exports) Bulgaria Czech Accession Latvia France Greece Italy Euro-zone Portugal Spain Rise in Euro Share of Invoicing Decline in Dollar Share of Invoicing Source: Author s calculations using ata from the ECB report an local country sources (etails in appenix) These changes in invoicing patterns may be consistent with a number of complementary hypotheses. First, there may be an increase prominence of the euro area an rest of Europe, or a ecline in the Unite States or ollar bloc countries, as a estination for exports. Secon, the increase in euro use an relate ecline in ollar use may be because accession countries have reuce the share of commoity type goos in their exports. Thir, these changes may be riven by proucer optimization uner changing covariance structures in macroeconomic fluctuations. Alternatively, there may have been a switch in market invoicing behavior from use of ollars to use of euros on the same proucts, with the same partners. 9 This change in behavior might occur because of a change over time in macroeconomic correlations, highlighte in equations (5) (7), with such changes possibly inuce by shifts in exchange rate regimes. By aressing these 9 Another, more munane explanation is that these results are purely ue to translation effects from changes in the ollar-euro exchange rate between 2000 an The valuation effects ue to the strong ollar uring 2000 mae the ollar value of exports isproportionately high for euro area countries. If the invoicing ata are base on nominal values, not real quantities, the ecline in the ollar against the euro through 2002 coul reuce the measure ollar invoicing share, even if actual invoicing patterns were unchange. The ollar appreciate by 3.1 percent from 2000 to 2001 an epreciate by 5.6 percent in The cumulative change from 2000 to 2002 was a ollar epreciation of 2.3 percent, much smaller than the total average eclines in ollar invoicing of exports of 16 percent for accession countries, an 14 percent for euro area countries. 15

18 hypotheses the work provies perspective on whether accession countries are invoicing as preicte by the theory an implie exporter optimization, or are potentially exposing themselves to excessive profit volatility an lower expecte profits. Destinations for Accession Country Exports. Di euro use increase in accession country trae because of increasingly close trae relationships with countries tie to the euro? Chart 4 provies ata on euro share in invoicing exports versus euro area share in total accession country exports for the years 2000 (inicate with lighter points) an 2003 (arker points). If all euro area trae was invoice in euros, an only euro area trae was invoice in euros, the ata points of this chart woul lie along the 45 egree line. The proximity of the three country-ata points for 2000 to the 45 egree line inicate that initial use of euros in invoicing roughly matche shares of the euro zone countries in exports for accession countries in that year. Yet by 2003 use of the euro in invoicing accession country exports far exceee the expane share of the euro area in country exports. All accession countries, with the single exception of Lithuania, ha euros play a larger role in export invoicing than woul be expecte purely ue to trae with countries within the euro area (ignoring, at this point, the issue of the composition of trae, which shoul reuce euro use even within the area to the extent that other currencies are use in invoicing homogeneous commoities an goos). As reflecte in istance from the 45% line, chart 4 shows that measure increases in euro invoicing between 2000 an 2003 well exceee the mil increases observe in accession country exports to the euro area. 16

19 Chart 4: Export Invoicing in Euros v. Exports to the Euro Zone, 2000 an 2003 Share of Exports to euro zone countries (in %) BULGARIA LATVIA Lithuania Cyprus CZECH REPUBLIC Czech Republic Polan Slovakia Bulgaria Estonia Latvia Hungary Slovenia Share of Exports Invoice in euros (in %) Relationship in 2000 Relationship in 2003 Another potential explanation is that accession country exporters are increasingly using euros to invoice exports to countries outsie of the euro zone, for example to the rest of the European Union or to countries tie to the euro through exchange rate arrangements. Some of this change may be attributable to changes in the exchange rate or currency orientation of traepartner countries. Such changes might inuce changes in the structure of covariances entering the invoice currency selection criteria, or might even inuce shifts in the hering currency equilibria for a particular type of goo. Within these partner countries, a key relate question is whether the competitors to accession country exporters are largely invoicing in euros or, for example, ollars. 10 Tables 2a an 2b provie etails on the concentration of accession country trae with European markets, the Unite States, an other countries heavily using the euro or the ollar. Table 2a shows that in 2003, the euro area accounts for between one quarter an sixty percent of accession exports. Other euro bloc countries are not big export estinations. Much more 10 Inee, for perhaps similar reasons that the accession countries choose to invoice trae largely in euros, many countries outsie of Europe choose to invoice their exports largely in U.S. ollars. As an example of this, Golberg an Tille (2005) show that among Australia, Japan, Korea, Malaysia, an Thailan ollar invoicing averages 73% for imports an 75% for exports. 17

20 influential are exports to the rest of the countries in the European Union but outsie the euro area. Malta is a consistent outlier, but otherwise these countries collectively account for close to three quarters of accession exports. 11 Table 2b shows that accession countries export much less to the Unite States an other countries with exchange regimes tie to the ollar, both in East Asia 12 an other regions 13. The Unite States purchases less than five percent of euro area exports of goos, except for Malta an Latvia. Accession countries also have low irect export links with other ollar bloc countries. Table 2a: Accession Country Exports to the Euro Bloc an the rest of the European Union, 2003 Country Euro Area Other Euro Bloc non-euro Area European Union Total Euro Bloc an EU Bulgaria Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Polan Slovakia Slovenia Non-euro area European countries are Denmark, Sween, the UK, an the 10 accession countries. As ocumente in Paoa-Schioppa (2004), Other euro bloc countries are countries with an exchange rate policy of pegging to the euro specifically Bosnia-Herzegovina, Montenegro, Serbia, New Caleonia, Benin, Burkina Faso, Cameroon, Cape Vere, Central African Republic, Cha, Comoros, Côte Ivoire, Equatorial Guinea, Gabon, Guinea-Bissau, Mali, Niger, Republic of Congo, Senegal, an Togo. 12 Camboia, China, Hong Kong, Inonesia, Korea, Malaysia, Philippines, Thailan, Vietnam. 13 We efine Other Dollar Bloc as: Australia, New Zealan, Canaa, Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuaor, El Salvaor, Guatemala, Honuras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, Venezuela 18

21 Table 2b: Accession Country Exports to the Unite States an the Dollar Bloc, 2003 Country Unite States East Asia Other Dollar Bloc Total Dollar Bloc Bulgaria Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Polan Slovakia Slovenia Source: International Monetary Fun, Direction of Trae Statistics Available ata oes not ifferentiate invoicing patterns as accession country exports to the euro area versus exports to other Europe an euro-bloc countries. Speculatively, given the ominance of trae with the rest of Europe in accession exports, these might also be the markets where accession countries are invoicing in euros. Consier the following hypothetical invoicing. Suppose that 100 percent of accession country trae with the euro area is invoice in euros (an overstatement given the composition of this trae across ifferentiate versus homogeneous an commoity-type goos) an 100 percent of accession country trae with the Unite States is invoice in ollars. The implie use of euros on accession country exports to other countries are shown in Chart 5, with these resiual exports primarily irecte at the rest of Europe. 14 These computations imply that euro invoicing occurs on an average of 37 percent of accession exports to countries outsie of the euro area an the Unite States. The variation across countries is large. Lithuania has zero implie euro invoicing on transactions outsie the euro area, while Hungary an Slovenia have euro invoicing have shares exceeing 70 percent on export transactions irecte outsie the Unite States an euro area. 14 The share of euro use in extra-euro area exports is constructe as: (share of total exports invoice in euros share of total exports sent to euro area) / (100-share of total exports sent to the euro area an the Unite States), where all shares are in percent. 19

22 Such statistics can be compare with invoicing patterns of countries alreay within the euro area, where the euro serves as the proucer currency as well as a potential vehicle currency elsewhere. Starting with ata on extra-euro area trae, euros are use in invoicing approximately 50 percent of extra euro-area exports. Uner the assumption that trae with the Unite States is exclusively in ollars, the computation implies that euros are use in invoicing nearly 60 percent of the remaining exports. The range is from a low of 30 percent for Greece to about 75 percent for Germany Chart 5: Euro Invoicing share in Exports to countries other than the euro zone an Unite States, 2003* %share of Total Extra-Euro Zone Exports Hungary Estonia Latvia Bulgaria Czech Cyprus Polan Slovakia Slovenia Germany Belgium Italy Spain France Portugal Luxembourg Greece 0.00 Accession Lithuania Euro Zone * * With Accession country shares estimate assuming that 100% of exports to the euro zone are invoice in euros an 100% of exports to the U.S. are invoice in ollars. Slovakia ata for Lithuania ajuste to zero from an estimate share of -7.4%. The Composition of Accession Country Exports. The previous section focuse exclusively on estination markets, without taking into account the composition of trae. Recall that the theoretical moel preicts a ominant role of hering in invoice currency choice for proucers whose goos face high elasticities of substitution in export markets. To highlight this point, this section categorizes accession country exports accoring to whether they are ifferentiate N (as are many manufacture goos), have uniform prices reference in inustry perioicals R (use for uniform goos not wiely trae enough to have a worl market, such as paper), or are consiere Walrasian W, which are homogeneous goos, mainly commoities such as ore with worl market prices, typically quote in a single currency an trae on organize exchanges. 20

23 Box 1 presents examples of Walrasian an reference price goos, with specific reference to exports of accession countries. For this construction of export composition shares, I use Rauch (1999) inices, which classify inustries into N, R, or W groups, an apply these inices to sort country-specific 4-igit SITC ata on exports for each accession country. Table 3 presents the resulting shares of ifferentiate (N), reference price (R), an organize exchange trae (W) goos in each country s exports. Table 3 The Composition of Accession Country Exports in 2003, by Pricing Metho Country Bulgaria Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta* Polan Slovakia Slovenia Differentiate Reference- Organize Total R+W share N Price R Exchange W * Source: Trae ata from UN Comtrae, an author s calculations. Malta ata from Differentiate proucts account for 62 to 83 percent of accession country exports. Organize exchange trae goos, often price in ollars, are typically a small proportion of the remaining exports an are generally less than 8 percent of exports. 15 Reference price goos are between 15 percent an 30 percent of accession country exports, with shares for Cyprus, Estonia, Latvia, an Lithuania all above 25 percent. All of the accession countries ha reuce shares of the more 15 Differentiate proucts comprise about three quarters of 2003 imports. 21

24 homogeneous goos (the R+W share) between 2000 an Declines were large for Cyprus an Lithuania. Taken together the shares of reference-price an organize exchange trae goos represent between 17 an 35 percent of accession country exports. While reuctions in these shares since 2000 may have accounte for some of the ecline in ollar use in invoicing exports of accession countries, R+W goos still represent a large portion of accession country exports. For the most part, the observe eclines are too small to explain shifts away from ollars in accession country export invoicing. Recall that the evience on invoicing changes between 2000 an 2003 was only available for Bulgaria, Czech Republic, an Latvia. Accounting for the changing share of trae with the euro area, Latvia an Bulgaria ha larger increases in invoicing in euros, while the Czech Republic ha relatively small changes. This pattern is not matche by R+W share eclines across these three countries, where the Czech Republic ha the largest change, Latvia some change, an Bulgaria very little change in composition of exports. If these R+W goos are invoice in ollars in European markets, the moel woul suggest that many accession countries uner-utilize ollars in invoicing exports in In some cases, the share of ollars use in invoicing total country exports is below the share of R+W goos in the export basket (Czech Republic, Estonia, Hungary, Latvia, Slovakia, Slovenia). Even though accession country exporters conuct much of their trae with other European countries, they still compete with proucers from aroun the worl, many of whom are likely invoicing these types of goos in ollars. The examples of pricing of copper, aluminum an paper pulp, shown in Box 1, illustrate the pervasiveness of ollar pricing on some proucts in these categories. If the accession country exporters are in fact invoicing less of their R+W type goos in ollars an instea invoicing these goos in euros, they may be exposing themselves to excess profit risk uner circumstances of movements of the euro-ollar exchange rate. 22

25 Box 1 Sample inustry profiles for reference price an worl market price goos Unwrought Copper: a worl market price goo. Rauch classifies commoity 6821, Copper an copper alloys, refine or not, unwrought (SITC rev igit classification), as a worl price goo. Accession countries export large quantities of unwrought copper, $942 million in 2003 (UN Comtrae), representing 0.5 percent of total Accession country exports. The main worl market for unwrought copper is the Lonon Metal Exchange (LME), with inustry profiles an reports referencing the LME prices. The official prices quote by the LME are in U.S. ollars per ton. The LME also traes unwrought aluminium, which also is a major export for the accession countries ($740 million in 2003), suggesting potential similarities in pricing an invoicing across both commoity categories. The largest accession copper exporters are Bulgaria an Polan (unwrought copper comprises 6.0% of Bulgaria s total exports by value in 2003). Almost all of Bulgaria s copper is smelte at Pirop, which is owne by Umicore, a Belgian company. The smelt copper is then exporte to Umicore s heaquarters in Belgium to be refine. So, Bulgaria s export prices are transfer prices not market prices, subject to qualification because they represent transfers within a corporation. Polan s main copper proucer is KGHM Polska Miez, which supplies 6% of the worl s copper accoring to AME Mineral Economics. KGHM posts a lot of information about its pricing structure on its website. They base their price on the LME an a a proucers premium which is base on the annual price announcements of Coelco (the biggest worl copper proucer), which are also mae in ollars. KGHM reports that the vast majority of copper sales are base on annual contracts where buyers agree to buy a certain tonnage a month whatever the market conitions, then pay each month base on the average market price over that month. A small share of sales is mae with spot contracts to eal with unexpecte shifts in supply or eman. Paper: a reference price goo. Paper is another major export for a number of the accession countries, incluing Estonia (2.5 percent of total exports in 2003), Polan (2.1 percent), Slovakia (2.0 percent), an Slovenia (1.8 percent). Pricing information is a little vaguer for this inustry because, by efinition, there is not an open worl market with frequently publishe price quotes. Inustry publications, such as Paperloop an Pulp & Paper Week, list monthly or quarterly market prices for various graes of paper, with these prices usually only mae available to subscribers. These perioicals publish prices for specific markets. Newsprint an pulp, both W -type goos, only have worl markets liste, but the inustry publications list printing an writing paper prices separately for North American, European, an sometimes Asian markets. When specific prices are mentione, Asian markets an North American markets were quote in U.S. ollars an Europe markets were quote in euros. One publication liste a full table of prices from FOEX (Finnish Options Exchange) which were all in euros. Lonon Metal Exchange: AME Mineral Economics: KGHM Polska Miez: Paperloop: Pulp & Paper Week: 23