How does Economic development in Eastern Europe affect Austria's regions? A multiregional general equilibrium framework

Transcription

1 How does Economc development n Eastern Europe affect Austra's regons? A multregonal general equlbrum framework by Johannes Bröcker and Martn Schneder ABSTRACT: Ths paper presents a framework for analysng the effects of economc development n Eastern Europe on Austra's regons. Therefore we concentrate on the effects of enhanced East-West trade, whch results from the economc development n the transton countres. The analytcal framework we use s a multregonal computable general equlbrum model for the Austran economy. The model s based on an Arrow-Debreu-equlbrum. It contans the 9 Austran Federal Provnces (NUTS-II) whch are lnked by trade flows wth 4 of the Central and Eastern European countres and wth the rest of the world. The smulaton results of one trade scenaro show how structural and welfare effects dffer for the Austran regons. JEL classfcaton: C68, F14, F15, F17, R13 Dresden Unversty of Technology, Germany; malto: broecker@rcs.urz.tu-dresden.de. Unversty of Economcs and Busness Admnstraton, Venna, Austra; malto: martn@wgeo1.wu-wen.ac.at. 1

2 1. INTRODUCTION The collapse of communsm was the startng shot of a fundamental change n the economc and poltcal structure of Europe whch hopefully wll end up wth the transformaton of the Central and Eastern European Countres (CEECs) to hghly developed market economes well ntegrated nto the world economy. Three processes are gong on wth dfferng speeds of progress n dfferent countres: the transton to a market economy and the reform of the nsttutonal framework, the catchng-up of the CEECs whch s expected to be the long term consequence of the reform, and the economc ntegraton of these countres nto the European and world economy. These processes wll lead to substantal changes of the volume and the composton of trade between the CEECs and the European Unon. The fact that an ncreasng trade volume s benefcal for both sdes has been confrmed by many studes (see Gasorek et al. (1992), Haaland and Wooton (1992) Sheehy (1995), Baldwn et al. (1997), for example) and s rarely dsputed. On the other hand t s often dscussed that the ncreasng trade volume could cause substantal dsrupton n some sectors or n some regons of the European Unon. Ths argument s true not only for the member countres of the EU, but also for the regons wthn member countres. The argument usually put forward s that CEECs have a tremendous labour cost advantage over EU members. That means, once they have learnt to apply recent technologes and have accumulated or mported the requred captal stock, they wll push western producers away. Ths wll be a problem for regons specalsng on labour ntensve sectors, and regons closer to the eastern border of today s EU wll be affected more than those further away. Whle some regons mght be compensated by attractng new demand from CEECs rasng ther ncomes, others could suffer from a lack of compettveness n the respectve export ndustry. One fault n ths argument s that low wages are assumed to persst forever. Wages, however, are not exogenous, but wll rse to the extent, that CEESs catch up wth respect to ther human and real captal stocks per unt of labour. A second fault n the argument s the neglect of general equlbrum repercussons. Even f we abstract from ncreased demand for western exports, a cheaper supply of goods from CEECs s not only a compettve threat, but at the same tme t gves the western regons the opportunty of ncreasng ther real ncomes due to more favourable terms of trade. Rsng real ncomes nduce addtonal demand, whch could substtute markets lost to compettors n CEECs. Furthermore, cheaper nputs mported from CEECs could strengthen the compettve poston of frms n EU regons. The am of ths paper s to pont to the mportance of these general equlbrum feedbacks by smulatng welfare effects of ncreasng trade flows from CEECs to the EU for the regons of Austra. We have chosen Austra because t s affected much stronger by the developments n the CEECs than any other member of the European Unon. Austra has a common border wth four of the CEECs, whch appled for membershp (Czech Republc, Slovaka, Hungary, and 2

3 Slovena, called ``CEEC-4 n the followng). We study a scenaro whch, at a frst sght, seems to be extremely threatenng to certan regons and sectors n Austra, namely an exogenous rghtward shft of the supply curve of all goods, delvered from CEEC-4 to Austra. It turns out that, even though any effect from ncreasng demand from CEECs s neglected, the scenaro would be benefcal for all Austran regons, even though to a dfferent degree. A computable general equlbrum model for the Austran economy s developed n ths paper. It s shown that the model s well capable of takng the relevant general equlbrum repercussons nto account. We start wth a bref dscusson of the macro-economc development of the CEEC-4 and CEEC-4/EU trade relatons n chapter 2, n order to motvate the comparatve statc smulaton exercse to be presented. In chapter 3 we wll take a bref look at the methods used for measurng welfare effects of changng trade regmes. The model wll be presented n chapter 4. After presentng the model, we wll demonstrate the welfare and producton effects our trade scenaro has on Austra s regons (chapter 5). Fnally, we wll draw some conclusons from the smulaton results and dscuss the tractablty and restrctons of our approach (chapter 6). 2. MACRO-ECONOMIC DEVELOPMENT AND TRADE OF THE CEEC-4 The Eastern Enlargement of the European Unon The CEEC-4 have already submtted ther applcaton for membershp of the European Unon. The last one was Slovena n Three of them (Czech Republc, Hungary and Slovena) belong to the group of sx countres whch the commsson recommended n July 1997 that accesson negotatons should begn wth. Although the CEECs are no members of the EU today, trade relatons between the EU the CEECs are almost lberalzed today. They were frst based on trade and economc cooperaton agreements, unlateral trade concessons, and more recently on assocaton agreements, known as 'Europe Agreements' (European Commsson, 1997). The macro-economc development n the CEEC-4 The system transformaton of the CEEC-4 s now at a farly advanced level. Havlk et al. (1997) summarsed that the major framework condtons for a well-functonng market economy have been establshed n the neghbourng countres. The transformaton crss has been overcome, a relatve strong economc growth has perssted, and the unemployment rates are decreasng. Monetary and fscal stablsaton programs have been successful. The rate of nflaton declned, the exchange rates are farly stable, the currences are convertble. There has been a strong declne of subsdes to prvate frms, prces are almost free from government control. Nevertheless, some dffcultes may persst and throwbacks are possble, manly for the followng reasons: dependence on cyclcal ups and down of Western European economes, nternal and external mbalances, structural defcts, balance of trade defcts (Havlk et al., 1997). The negatve trade balance s a common problem of all transton countres. Ths s especally true for the Czech Republc. The Czech economy suffered from a severe fallback n 3

4 1997. The GDP growth fell snce 1995, where t has reached ts peak (5.9 %). Podkamner et al. (1998) estmated a growth rate for 1997 of only 1.3 %. The man reason for that setback s that the Czech authortes tred to keep the nomnal exchange rate constant snce Rates of nflaton hgher than n Western Europe gradually undermned the compettveness of Czech companes on foregn markets. So exports stagnated, whle mports were boomng. As a consequence of that, the current account defct expanded dramatcally (see Pöschl, 1998, p. 32). Besdes other structural problems, the captal markets are stll underdeveloped (see Bruss and Ochmann, 1996, pp ). Structural reforms n ths area stll have a long way to go. Trade relatons between the CEEC-4 and the European Unon One of the man characterstcs of the CEEC-4 trade before 1989 was ts 'dual' structure. The trade wth other countres of the Councl for Mutual Economc Assstance (CMEA) had the characterstcs of advanced economes (.e. sellng and buyng ndustrally advanced products to/from each other nvolvng a consderable amount of ntra-ndustry trade) whle trade relatons wth the West had (and to a certan extent stll have) more of the characterstcs of the trade of developng countres (.e. exportng relatvely raw-materal-ntensve products and mportng ndustrally more processed, hgher value added products) (Landesmann and Székely, 1995, p. 43). So East-West trade followed the classcal Heckscher-Ohln pattern, where trade flows are determned by dfferent factor ntenstes 1. The development of trade between the CEECs and the EU after 1989 can be characterzed by three man phases. From 1989 to 1992 ('early restructurng phase') the economes of the CEEC- 4 suffered from the transton recesson. Output and ncomes declned. Beyond trade creaton, there has been a shft n the geographcal pattern of trade away from ther former trade partners to the EU 2 (trade redrecton). The resultng East-West trade patterns were domnated by nterndustry trade. From 1992 to 1994/95 the CEEC-4 economes began to recover ('recovery phase'). The volume of East-West trade has grown very rapdly. Besdes nter-ndustry trade, ntra-ndustry trade began to rse. Regardng the sectoral structure, the transton perod ( /95) was characterzed by substantal changes n the mport structure of the CEEC-4 and only modest varatons n the composton of ther exports (many studes obtaned ths result; see Halpern (1995), Havlk (1995), for example). The shft of the CEECs mport structure s characterzed by a decrease of mports of ntermedate and nvestment goods and an ncrease of consumpton goods mports (Fan and Portes, 1995, p. 10). On the export sde, there s a close relaton between the commodty structure of the CEECs exports and ther ndustral structure. So the modest changes of the export structure reflect a lack of ndustral restructurng 3. The 1 Ths fact s not undsputed. See Agnger et al. (1994) and Landesmann and Székely (1995) for the explanatory power of the Heckscher-Ohln framework for explanng East/West trade. 2 Many studes confrmed ths result. See Halpern (1995, p.61), for example. 3 Mtov (1996, p. 166) explans ths development by the followng factors. There s a shortage of nvestment captal, especally for technologcal development. Inapproprate prcng structures (low prces of some prmary nputs and utltes) create dsncentves to restructure or ratonalze the resource allocaton. Whle ths may be benefcal n the short term, such a strategy s dsastrous n the longer run. Foregn nvestment s stll lmted n 4

5 CEECs exports are strongly based away from captal-, R&D- and skll-ntensve branches and towards energy-ntensve and labor-ntensve branches. These specalzaton patterns have been strengthened snce 1989, but there are a lot of exceptons and the development dffers greatly between countres (Landesmann, 1995, p. 19). Durng the 'catchng-up phase' (1994/ /20) the mport demand of the CEEC-4 gves Austran frms the possblty to export nvestment goods, hgh-qualty consumpton goods and servces (fnancal servces, for example). On the other sde, the expanson of the export capactes of the CEEC-4 wll put a pressure on producers of resource-ntensve commodtes, labour-ntensve commodtes (leather and clothes) and partly even captal-ntensve commodtes (steel and coal) n Austra (Holzmann and Zukowska-Gagelmann, 1996, p. 194). As the CEEC may close ther technologcal gap and become more and more smlar to the Western European countres, the share of ntra-ndustry trade wll rse and the share of nter-ndustry trade wll fall (Landesmann, 1995, p. 19). We wll now take a short look at the volume of future East-West trade. Most of the studes whch try to project the volume of future East-West trade are based on the 'gravty approach to trade', where the volume of trade depends on the dstance between the two tradng countres as well as ther ncomes and ther sze. Gravty models are able to estmate the total volume and the geographcal structure of East-West trade, but not ther sectoral structure 4. We have surveyed three studes whch are based on the gravty approach (Baldwn (1994), Holzmann et al. (1993), Holzmann and Zukowska-Gagelmann (1996)). They all yeld smlar results. Frst, the trade volume between the Western and Eastern European countres should ncrease several-fold. The exports of the CEECs to Western Europe of the projecton year are expected to be two to three tmes the exports of the base year. The mports of the CEECs from Western Europe are expected to be two to fve tmes the mports of the base year. The dfferent results depend on dfferent reference and projecton years, dfferent country groups, and dfferent assumptons underlyng the scenaros. Second, trade wth the CEECs should not lead to trade defcts for the Western European countres,.e. the projectons for Western European countres exports are typcally hgher than for ther mports. So Western European countres should not suffer from trade mbalances. A trade scenaro In our formal framework, the regonal mpacts of changng trade flows between Austra and ts eastern neghbours s smulated by exogenous shfts n CEEC-4 s supply and demand curves for goods traded wth Austra. General equlbrum effects are only taken nto account for Austra, whle the nternal structure of CEEC-4 s not explctly modelled. sze and concentrated mostly n the tradtonal areas. There s a general delay n the structural reforms n all CEECs. 4 It s sometmes argued that the theoretcal foundaton of the gravty model s weak. 5

6 It s not possble, however, to catch all these processes by a sngle smulaton scenaro. Dfferent shfts n dfferent sectors have to be ntroduced n order to smulate the effects charactersng dfferent perods of transformaton, catchng up and ntegraton. Lmtaton of space prevents us from any attempt to develop realstc scenaros for all phases of the process. Our man purpose n ths paper s to demonstrate that takng general equlbrum feedbacks nto account s mportant and tractable. It s mportant, because the overall results mght dramatcally dffer from what one expects from smplstc partal equlbrum consderatons. The trade scenaro smulated n ths study s chosen such that ths fact s made as clear as possble. The nflow of cheaper mports from CEECs n all sectors, where they have been strong already n the past, s usually regarded as most threatenng for certan regons and sectors n Austra, and hence the most problematc process under a regonal equty pont of vew. Therefore our scenaro solates ths sngle component of the entre process. It s assumed that the CEEC-4 wll ncrease ther export capactes of all sectors to the three-fold level of For smulatng ths scenaro n our model, we have to shft the CEEC-4s mport supply functons for every ndustry to the rght 6. That means that the CEEC-4 are wllng to supply a hgher quantty of exports at a gven prce. In order to make such an ncrease possble, the CEEC-4 have to buld up new producton capactes. So ths scenaro reflects a successful restructurng of ther ndustres. 3. MEASURING THE WELFARE IMPLICATIONS OF CHANGING TRADE REGIMES The two major methods to evaluate the economc benefts of economc ntegraton are econometrc evaluatons and computable general equlbrum models. Econometrc evaluatons estmate the parameters of usually smple theoretcal models (or even from a sngle equaton) by usng econometrc technques. They usually focus on questons as the estmaton of trade creaton and trade dverson effects or growth effects of ntegraton (Baldwn and Venables, 1995, p. 1626). The drawback of econometrc evaluatons s that they cannot capture the complcated nterplay of effects whch may be mportant for massve polcy changes such as the ntegraton of the CEECs nto the European Unon. Computable general equlbrum models are the predomnant framework for analyzng the economc benefts of economc ntegraton. Baldwn and Venables (1995, p ) classfed appled CGE models nto three generatons, whch correspond to three types of welfare effects. The models of the frst generaton are statc and assume perfect competton. Gans from trade n these models are due to mproved effcency from the reallocaton of resources among sectors n dfferent countres. The models of the second generaton are also statc, but allow for ncreasng returns to scale. Trade lberalzaton enables mproved effcency through scale 5 Ths magntude s n lne wth the results of many studes whch predcted the future volume of trade between the Eastern and Western European countres by a gravty model (see chapter 2). Although these studes predct only the total volume of trade n the long-run (wth partal or total ncome catch-up of the Eastern European countres), they can serve as a gudelne for the magntude. 6 Remember that mport supply s CEECs supply of goods mported by Austra from these countres, accordng to our termnology. 6

7 economes and mproved competton. Thrd generaton models are dynamc and nclude accumulaton effects. Savngs wll create addtonal nvestment and therefore ncrease the captal stock. Note whle the welfare gans n the frst two generatons of models reflect ncreased effcency for a gven amount of nputs (statc allocaton effects), thrd generaton models augment the captal stock and therefore ncrease the amount of avalable nputs (Cooper, 1996, p. 256). Although CGE models suffer from some weaknesses (see Gelauff and Graafland, 1994, for example), they are the most approprated tool for analyzng benefts from economc ntegraton. 4. THE MODEL 4.1 Summary The model to be mplemented s a statc perfect competton computable equlbrum model wth - 11 regons, 9 domestc regons plus 2 foregn regons (CEEC-4 and ROW), - 10 producton sectors, - 2 factors of producton (labor and captal), and - a household sector representng all types of domestc fnal demand, ncludng nvestments. There s no separate publc sector,.e. publc servces are contaned n the producton sectors and publc ncomes and expendtures are part of the representatve household s ncomes and expendtures. For the delneaton of regons and sectors see the appendx. The model s dentcal to the one descrbed by Bröcker (1998). The only modfcaton s to ntroduce foregn trade relatons. Bröcker s (1998) model s closed. For detals the reader s referred to the cted paper. We follow ts notaton as close as possble. The domestc economy s modeled as a general equlbrum, whle the foregn regons are only partally represented by mport supply and export demand 7 functons. Hence, for each domestc sector (frms and households) all returns and expendtures are covered by the model, whle only a subset of flows nto and out of foregn regons appears n the model. A producton sector n a domestc regon s represented by a prce takng proft maxmzng frm, producng a sngle output by ntermedate nputs and factor nputs under constant returns to scale. Therefore ts output prce equals mnmum unt cost. Intermedate nputs are bought n the own regon, n other domestc regons or abroad. As usual n multregonal computable equlbrum models, products of one sector from dfferent regons of orgn are regarded as dstnct, and the degree of substtutablty between them s lmted. Outputs are sold for domestc ntermedate or fnal use, or they are exported. Labor and captal are n fxed supply n each 7 Import supply means a foregn regon s supply of goods mported by domestc frms or households. Smlarly, export demand means foregn demand for domestcally produced goods. 7

8 domestc regon. Regonal factor returns flow to a regonal representatve household, maxmzng homothetc utlty by expendng the ncome for goods, bought from domestc producers or mported. All preferences and technologes have a nested constant elastcty of substtuton (NCES) form. The mport supply and export demand functons are constant elastcty functons. Thus far, the specfcaton s standard. The usual procedure would be now to set up a complete socal accountng matrx, to calbrate the free parameters of preferences and technologes and to smulate counterfactual scenaros. The necessary nformaton, however, s usually not avalable. For a standard calbraton we would need a full nterregonal nput-output table reportng, for each sector n each regon, all nputs by sector and regon of orgn. Ths s sometmes called an deal nput-output table. As nobody ever had such a table, t s usually produced by a data generatng procedure based on ad hoc assumptons. Authors do not lke to uncover the recpes of the wtchcraft practced n ths stage. Standard ngredents are rules of proportonalty, constant ratos, RAS-procedures and gravty equatons. In contrast to these mxtures, our phlosophy n ths context s smple and clear: don t use data, whch you don t have. Or puttng t n a postve way: Only allow for free parameters whch can be calbrated by exstng nformaton. The nformaton we do have s - a natonal nput-output table, - nformaton on the locaton of sectors (such as employment by sector and regon), - nformaton on regonal factor prces, and - nterregonal transport cost. What we don t have, however, s - nput-output nformaton on the regonal level, and - nterregonal trade data. Thus we have to ntroduce assumptons nto our theoretcal model (not nto the data generatng process) makng t possble to dspense wth the latter type of nformaton. Three assumptons lead to a soluton: - Frst, producton technologes of frms and household preferences do not depend on locaton. - Second, a poolng concept s appled, known as the Chenery-Moses approach n the nputoutput lterature (Batten and Boyce, 1986). Accordng to ths approach, all outputs of a sector orgnatng from dfferent regons (domestc and foregn) and used n some regon of destnaton are merged nto a sectoral pool, from where customers take goods for ntermedate or fnal use. The goods n the pool are called pool goods. They are composte goods generated by a NCES aggregator (called the pool-aggregator ), usng the goods from dfferent orgns as nputs. The regonal composton s obtaned from cost mnmzaton. Ths poolng approach amounts to the same thng as to assume that all utlty and producton functons contan nests on the lowest level aggregatng goods from dfferent regons (one nest for each nput sector), and that these nests do not dffer between users. Such an 8

9 assumpton s ndspensable, because we have data for calbratng the poston parameters for just one such nest per sector. The nformaton comes from the regonal dstrbuton of the respectve sector. - The thrd assumpton s that nterregonal trade s costly, wth costs dependng on nterregonal dstance. These assumptons mply that nput-output coeffcents n the benchmark equlbrum endogenously vary over regons n response to prce varatons, and that trade flows fulfll a gravty equaton. Unfortunately, the calbraton procedure becomes more complcated wth these assumptons. Parameters can not drectly be deduced from the data, but have to be found by solvng a farly large system of non-lnear equatons. The reader s referred to our cted paper (Bröcker, 1998) for detals. The model s calbrated wth 1994 data. Hence, the benchmark equlbrum reproduces the statstcal observatons of 1994 used for parameter calbraton. After a successful calbraton we are able to smulate counterfactual scenaros. Two types of equlbra wll be studed, an equlbrum wth complete market clearng and a fxed wage equlbrum. In the latter, goods markets and the captal markets clear, whle unemployment arses due to a fxed lower bound for real wages. 4.2 Formal descrpton We begn wth ndex notaton. Subscrpts r, s = 1,..., N, N + 1,..., R refer to regons. { 1,..., N } are domestc, { N R} Superscrpts, j = 1,..., I refer to sectors. +1,..., are foregn regons. Superscrpt k = 1,..., K refers to factors. Superscrpts are always understood to run through ther whole range, whle ranges of regonal subscrpts are always explctly gven. Let S r and P s denote supply and pool-good quanttes, respectvely, and let p r and q s denote ther respectve prces. For domestc regons S r s just the sectoral output x r, S = x, r = 1,..., N. For foregn regons S r s mport supply, whch s assumed to be an r r ncreasng functon of prce, S ( p ) r r r µ r = π. π r s a poston parameter to be calbrated, and µ r s an elastcty, whch may vary over regons and sectors. In the lmtng case µ r the prce p r s constant. 9

10 For domestc regons the quantty of pool goods s obtaned from ntermedate and fnal demand, P = d + a x, s = 1,... N. s s s j s j j d s s fnal demand resultng from utlty maxmzaton under the budget constrant k f w = d q, s = 1,..., N. s k s k s s f s k s the factor stock and w s k s ts prce. All regonal factor returns are expended completely n the same regon. There s no savng (remember that d s ncludes nvestment demand) and no nterregonal flow of funds. For foregn regons P s s export demand, whch s decreasng n ts prce, ( ) ε P = τ q, s = N + 1,... R, s s s s wth poston parameter τ s and elastcty ε s. We do not allow for the lmtng case ε s, because t would mply that all prces are determned from abroad and that regons would specalse completely, whch s unrealstc. The condton of goods market clearng completes the descrpton of the quantty system, S = t P, r = 1,..., R. r R rs s = 1 s t rs s the trade coeffcent, measurng the cost mnmzng nput of goods from r (ncludng mports) per unt of pool good n s. As to the prce system, we need the equatons for factor prces w r k, prces of pool-goods q s, and output prces pr, r = 1,..., N. Concernng factor prces, dfferent cases have to be consdered. The factor prce of captal s always obtaned from the market clearng condton, whle that for labor s obtaned ether as an exogenous lower bound for the wage rate or from market clearng. The former case apples n a fxed wage scenaro, f there s a bndng lower wage bound, the latter apples otherwse. These dfferent possbltes can be summarzed n a complementarty condton, f c x, w w, r k j r kj r j r k fr k cr kj xr j ( wr k wr k ) = 0, r = 1,..., N. j r k 10

11 w r k s the lower bound. It may be set equal to zero f one does not want t to be bndng. c r kj s the cost mnmzng factor nput per unt of sector j output n regon r. Output prces and prces of pool goods equal ther unt cost, r r j r j j p = a q + c w, r = 1,..., N, r k q = t p, s = 1,..., R. s R rs r = 1 r k r k The system s complete, f we specfy technologes and preferences, from whch nput coeffcents, trade coeffcents and fnal demand are derved. The specfcaton of technologes and calbraton of ther parameters are treated n the next subsecton. The dervaton of trade coeffcents, however, deserves a specal treatment here. As already mentoned, nterregonal trade s modeled by takng transport cost explctly nto account. The term transport cost s a short cut for all knds of dstance dependent costs related to nterregonal trade. For the sake of smplcty we adopt Samuelson s (1954) ceberg model. A good transported from r to the pool n s loses a certan share η of ts quantty per unt of dstance. Hence, f z rs denotes dstance from r to s, then the prce per unt of good from r arrvng n s (the c..f. prce ) s ( η rs ) v = p exp z. rs r Furthermore, let ct ( v s v 1 Rs ),..., be the unt cost functon derved from the pool aggregator. Arguments of the pool aggregator are quanttes avalable n the regon of destnaton, and arguments of the correspondng cost functon are the respectve c..f. prces. Note that ct has a sectoral ndex, but no regonal ndex. The aggregator s the same for all destnatons. 8 Accordng to Hotellng s lemma, the nput per unt of pool good s the frst dervatve of the unt cost functon. Therefore, the nput of goods at the place of orgn per unt of pool good n the place of destnaton s t rs ct = v rs ( η zrs ) exp. 4.3 Technology and preferences Households are assumed to have a one-level CES utlty, whch s dentcal for all regons. It s unquely specfed by the elastcty of substtuton σ H and an I-Vector of poston parameters. 9 8 There s one excepton: Inputs from foregn regons nto foregn pools are restrcted to zero. 9 The classcal reference for parametrsng NCES functons s Keller (1976). There the poston parameters are called dstrbuton parameters, and are normalzed such that they add up to unty. We dspense wth normalzaton n order to have a degree of freedom for freely choosng unts of all goods (see Bröcker, 1998, appendx). 11

12 Ths vector s calbrated such that the fnal consumpton vector n the natonal nput-output table s exactly reproduced by the correspondng natonal aggregate n the benchmark equlbrum of the model. Smlarly, technologes of frms are unquely specfed by the substtuton structure I + K -vector of poston parameters, whch s calbrated usng natonal (see fgure 1) and an ( ) nput-output data. Fgure 1: Representaton of the technology of the frms and the pool aggregators Frms Pool aggregators Source: The authors own draft. Fnally, the pool aggregators have a 2-level substtuton structure (see fgure 1) and an R- vector of poston parameters. Remember that the aggregators are the same for all regons of destnaton, except for the fact that foregn regons are not allowed to buy from foregn regons. Hence, the upper nest s mssng for foregn regons of destnaton. The poston parameter s calbrated such that, for each sector, the data on regonal employment and on mports by foregn regons of orgn are exactly reproduced by the benchmark equlbrum. 4.4 Elastctes and transport rates Only poston parameters are obtaned from calbraton. External nformaton s requred for fxng the remanng parameters. For elastctes we fx a base case (table 1) and vary the elastctes systematcally n a senstvty analyss. Though we have consulted an extensve body of econometrc lterature 10 a consderable degree of arbtrarness remans, because estmates vary 10 Sadoulet (1995), Ho and Jorgenson (1994), Brenton and Wnters (1992), Harrson et al. (1991), Jones and Whalley (1988), Fullerton et al. (1983), Mansur and Whalley (1984), Burnaux et al. (1990), Mensbrugghe et al. (1990). 12

13 over wde ranges. Note that we apply the small economy assumpton, as far as mport supply from the rest of world s concerned 11. Table 1: Elastctes and transport rates used n the model. Producton Transport Import demand Import supply Export demand Transport Type of elastcty Substtuton Substtuton Substtuton Prce Prce Prce Prce (per 1000 σ P σ T σ M µ CEEC µ ι ROW ε ι CEEC ε ι ROW η ι AGRICUL CLOTHES METAL TECH MINSPEC FOOD CONSTR TRADEACC SERVICE SERVICE Source: The authors own assumptons, based on a lterature survey. The transport rates shown n table 1 are crude guesses. Jontly wth the parameters σ T, these rates control the dstance senstvty of nterregonal trade. It can be shown that nterregonal trade n our model fulflls a gravty equaton wth dstance functon exp ( λ z rs ), wth λ = σ 1 η. Bröcker (1984, p. 131) estmated λ-parameters around 1 per thousand. ( T ) klometers for nternatonal commodty trade. Hence, for σ T n the range 5 to 8, η should be n the range of 1 7 to 1 4. Thus, at least the order of magntude of the gven rates for goods producng sectors seem plausble. Consderably hgher rates have to be assumed for constructon and servces, as the respectve output s less easly tradable. rates km) 5. SIMULATING THE EFFECTS OF SCENARIO 'CEEC EXPORT EXPANSION' Drvng forces wthn the model Before we turn to the results of our scenaro, we want to deal wth the drvng forces whch work nsde the model and determne the results. Ths gves us the opportunty to mprove our understandng of the behavor of the model and to avod the "black-box feelng", whch s often assocated wth general equlbrum models. 11 The small country assumpton states that a country faces fxed world market prces. They can be fxed ether on both mport and export sdes (ths case s also referred to as a "small open economy") or on the mport sde only ("'almost' small open economy"). We have fxed the mport prces for mports from the ROW by choosng a very hgh value for the mport supply elastcty (µ ι ROW). 13

14 The Austran demand for mports of a specfc sector depends on the relatve prce of the mports to domestcally produced commodtes. As long as ths relatve prce remans constant, mport demand remans constant. Snce mport supply of the CEEC-4 ncreases, a dsequlbrum n form of excess supply occurs on the respectve mport market. So the prce of the mports has to fall n order to clear the mport market. The fallng mport prce leads to a rse of Austran mport demand and to a fall of CEEC-4 mport supply. Ths leads to a new equlbrum on the mport market. The resultng market clearng amount of addtonal mports s smaller than the ntal amount of addtonal mport supply of the CEEC-4 and depends prmarly on the prce elastcty of the mport supply of the CEEC-4 (µ CEEC) and the elastcty of substtuton of the domestc pool aggregators between foregn and domestc products (σ M). The Austran demand for mports from the Rest of the World decreases because of the fallng prces for mports from the CEEC-4. So mports from the CEEC-4 are substtuted for mports from the ROW and for domestcally produced commodtes. Snce the demand for domestcally produced commodtes of the sectors wth the ncreasng mport penetraton declnes, a part of the producton factors becomes unemployed. So factor (and therefore commodty) prces have to fall n order to clear the markets. If we assume that the mport penetraton affects a labour-ntensve sector, employment n ths sector declnes. So the wage rate has to fall to clear the labour market. Ths causes the frms n other sectors to ncrease employment. So factors are reallocated between sector due to changes of ther relatve prces. The declne of the mport prces mproves the terms of trade 12 n Austra. It s a well-known statement n trade theory that a rse of the terms of trade ncreases a country's welfare (see Krugman and Obstfeld, 1994, p. 92, for example). Total demand for domestcally produced commodtes conssts of two components, namely fnal demand of the households and export demand of the foregn countres. Fnal demand of the domestc households rses because an ncrease of an country's welfare (whch s equvalent to an ncrease of real ncome) results n an ncrease of fnal demand. The commodty prces n Austra declne because of the fall of the factor prces and because of the fact that cheaper mports are used as ntermedates n producton. So demand of the foregn countres for Austran exports ncreases because of the fallng prce level n Austra 13. Ths damps the fallng prce level and helps to mprove the terms of trade. In the model specfcaton wth fxed real wages the causal chans are n prncple the same. The man dfference les n the functonng of the labour markets. In contrast to the flexble wage specfcaton, where only a reallocaton of resources between sectors takes place, total aggregated output and employment rses because of the nfntely elastc labour supply (up to a predetermned lmt). Welfare effects are hgher n the model specfcaton wth fxed real wages than n the model specfcaton wth flexble wages. To understand ths, we have to take a look at the ntal state of the economy before we smulate any scenaro (= benchmark equlbrum ). In 12 The terms of trade are defned as the rato of a country's export prces to ts mport prces. 13 Notce that the balance of payment condton s automatcally fulflled for each Austran regon due to the budget restrcton of the households. Therefore the balance of payment condton s fulflled for Austra as a whole. Ths mples that the value of the addtonal exports equals the value of the addtonal mports. 14

15 an economy wth flexble wages, the wage rate s obtaned from the market clearng condtons. Ths means that total labour supply s always employed. So there s an effcent use of all resources. In an economy wth fxed (real) wages unemployment can occur, f the lower bound for the wage rate s bndng. In ths case, the lower wage bound deters the frms to demand enough labour. As a consequence of ths, unemployment occurs. So fxed wages lead to dstortons wthn the economy. On the one hand, these dstortons lead to a lower welfare level of the economy n the benchmark equlbrum (compared to a non-dstorted economy,.e. flexble wages). On the other hand, the reducton of ths dstortons (e.g. by foregn trade) wll lead to hgher welfare gans because there are more "effcency reserves" wthn the economy 14. Employment n Austra s rsng due to the rsng mports from the CEEC-4 (table 3). Ths result seems to be mplausble, so we have to explan the forces whch lead to ths result. The common-sense argument s that addtonal mports wll hurt the respectve ndustry and wll lead to employment losses. Ths s true n a partal equlbrum context, where addtonal mports of one sector lead to more competton n ths sector. So employment n ths sector wll declne, snce demand shfts from domestcally produced commodtes to mported commodtes. In a general equlbrum context, ths wll happen too. But there are addtonal forces workng whch lead to an ncrease of employment. As we have ponted out above, the ncrease of the terms of trade due to the declnng mport prces leads to an ncrease of welfare n Austra, whch s equvalent to an ncrease on real ncome. Ths s the same effect as n the case of flexble wages. The only dfference s that labour markets react on excess demand by quantty rather than prce adjustment. So ether the wage rate (flexble wages) or employment (fxed wages) ncreases. Smulaton results Table 2 presents the sectoral results for both the flexble and the fxed real wage model specfcaton. What we can obtan from the results are the substtuton effects on the mport sde (mports from CEEC-4 for mports from ROW and for domestc producton) and the reallocaton of producton factors between sectors. In addton to these effects, ntersectoral nterdependences and preferences of the households nfluence the results. Wth flexble wages, total output remans constant. Wth fxed real wages, total output rses by 0.6 %. 14 In addton to the fxed real wage rate, we have to assume that there s enough unemployed labour force whch can be demanded by the frms. 15

16 Table 2: Sectoral results of scenaro 'CEEC Export Expanson': Imports, exports, producton and fnal demand (per cent changes compared wth the benchmark equlbrum) Flexble wages Imports Exports Producton Fnal demand CEEC ROW Sum CEEC ROW Sum AGRICUL 105.2% -3.8% 4.1% 1.3% 1.5% 1.5% -0.9% 0.3% CLOTHES 104.0% -5.3% 1.7% 1.6% 1.9% 1.9% 0.6% 0.6% METAL 169.2% -8.7% 10.5% 1.4% 1.9% 1.8% -2.7% 1.4% TECH 170.6% -3.5% 1.4% 1.3% 1.8% 1.8% 0.7% 0.4% MINSPEC 104.2% -5.0% 5.2% 1.6% 1.9% 1.9% -0.8% 0.5% FOOD 96.8% -2.4% -0.2% 1.1% 1.5% 1.5% 0.4% 0.2% CONSTR 121.0% -2.1% 9.6% 0.9% 1.5% 1.4% 0.2% 0.2% TRADEACC 121.2% -1.7% 4.5% 0.7% 1.2% 1.1% 0.1% 0.1% SERVICE % -1.5% -1.4% 0.6% 1.1% 1.1% 0.3% 0.1% SERVICE % -1.8% 4.5% 0.7% 1.2% 1.2% 0.1% 0.2% Sum 142.4% -4.3% 3.1% 1.3% 1.8% 1.7% 0.0% 0.3% Fxed real wages Imports Exports Producton Fnal demand CEEC ROW Sum CEEC ROW Sum AGRICUL 105.1% -3.8% 4.0% 1.6% 2.0% 1.9% -0.1% 0.9% CLOTHES 104.3% -4.9% 2.1% 2.0% 2.3% 2.3% 1.3% 1.1% METAL 169.2% -8.5% 10.7% 1.7% 2.3% 2.2% -2.0% 1.9% TECH 170.5% -3.2% 1.7% 1.6% 2.2% 2.1% 1.4% 0.9% MINSPEC 104.4% -4.8% 5.4% 1.8% 2.2% 2.1% -0.2% 1.0% FOOD 96.9% -2.3% -0.1% 1.3% 1.8% 1.7% 1.0% 0.7% CONSTR 121.0% -2.2% 9.5% 1.2% 1.9% 1.9% 0.8% 0.9% TRADEACC 121.3% -1.7% 4.6% 0.9% 1.5% 1.5% 0.7% 0.7% SERVICE % -1.5% -1.3% 0.9% 1.5% 1.5% 0.9% 0.7% SERVICE % -1.9% 4.3% 1.1% 1.8% 1.7% 0.8% 0.8% Sum 142.4% -4.1% 3.3% 1.6% 2.2% 2.1% 0.6% 0.9% Source: The authors own calculatons. Two factors prmarly cause dfferent results for the Austran regons: the sectoral composton of the ndustral structure of a regon and the dstance to the CEEC-4. The sectoral composton of a regon nfluences regonal results through the factor prces. In regons wth a hgher share of sectors whch suffer from the mport competton there s a greater pressure on the factor prces than n other regons. So there s an nverse relatonshp between structural adjustment needs and the rse of real ncome. Dstance matters because the nearer an Austran regon les to the CEEC-4, the hgher s the CEEC-4s share n the total volume of foregn trade of the regon 15. Due to nterregonal trade flows, supply as well as demand shocks rpple throughout the economy. Three measures are used to present the mpacts of ths scenaro on the dfferent Austran regons. The change of regonal real ncome measures how the welfare of the regonal households s affected. A structural change ndcator (s r ) measures the magntude of adjustment per regon. Although structural change s costless n the neoclasscal world, ths structural change ndcator can be nterpreted as a measure for the burden structural change has on a regon n realty. It s defned as the percentage change of the sum of the absolute values of the changes of x x, dvded by total regonal output ( x r denotes the output of sectoral output n a regon ( ) r sector n regon r n the counterfactual equlbrum, x r r s the correspondng output n the 15 Ths results from the gravty approach to trade. 16

17 benchmark equlbrum). The thrd measure s the ncrease of employment n the regons. Notce that employment rses n the fxed real wage specfcaton only. s r = x r x x r r * 100% Table 3: Regonal results of scenaro 'CEEC Export Expanson': real ncome, regonal structural change and employment (per cent changes compared wth the benchmark equlbrum) Flexble wages Fxed real wages Real Structural Employment Real Structural Employment ncome change ncome change Burgenland 0.28% 0.56% % 0.97% 1.01% Carntha 0.27% 0.47% % 0.93% 0.99% Lower Austra 0.27% 0.63% % 0.96% 0.94% Upper Austra 0.21% 0.64% % 0.91% 0.86% Salzburg 0.23% 0.45% % 0.87% 0.94% Styra 0.23% 0.62% % 0.89% 0.91% Trol 0.19% 0.50% % 0.83% 0.87% Vorarlberg 0.17% 0.57% % 0.90% 0.80% Venna 0.35% 0.25% % 0.92% 1.22% Sum 0.26% 0.49% 0.86% 0.91% 0.99% Source: The authors own calculatons. The welfare gans range from 0.17 % (Vorarlberg) to 0.35% (Venna) for flexble wages and from 0.66 % (Vorarlberg) to 1.05 % (Venna) for fxed real wages (see table 3). Whle total employment remans constant for flexble wages, an ncrease of employment wll occur for fxed real wages. The addtonal employment ranges from 0.80 % (Vorarlberg) to 1.22 % (Venna). We can make two observatons regardng the structural adjustment needs. Regonal structural adjustment needs are lower (compared to the regonal welfare gans) and more equally dstrbuted n the fxed real wage specfcaton than n the flexble wage specfcaton. The reason for that s that n the latter model specfcaton, the only way to ensure a new equlbrum n case of a dsequlbrum s resource reallocaton between sectors due to changes n relatve prces. The less flexble the prces are, the less resource reallocaton between sectors can take place. In the case of fxed real wages, only captal s reallocated between sectors, whereas adjustment on the labour markets s done by quantty rather than prce adjustment. So less restructurng takes place. Senstvty analyss Results from appled general equlbrum models are often presented as a sngle soluton of a determnstc system. But there are several features of these models whch gve rse to uncertanty concernng the relablty of the soluton. Mensbrugghe et al. (1990, p. 174) dvded these uncertantes n three broad groups, namely uncertanty about the model specfcaton, the macro- 17

18 closure and the values of parameters such as elastctes, transport rates, etc. Though we have consulted an extensve body of econometrc lterature, a consderable degree of arbtrarness remans, because estmates vary over wde ranges. Especally the values for the sectoral transport rates are only crude guesses. So we wll perform extensve senstvty analyss for evaluatng these uncertantes. In addton to ths, another mportant aspect of senstvty analyss s that t wll mprove our understandng of the behavor of the model. The level of natonal real ncome crucally depends on the value of three elastctes: mport ε s (see supply elastcty ( µ r ), mport demand elastcty, and export demand elastcty ( ) fgure 2 16 ). If the mport supply elastcty of the CEEC-4 s hgh, a small fall of the mport prce leads to a hgh fall of foregn mport supply. What follows s that a small fall of the mport prces s suffcent to clear the mport markets. The less the mport prces are fallng, the smaller s the welfare effect. The same argumentaton can be used to explan the senstvty of the welfare gans to the domestc mport demand elastcty. The hgher the elastcty, the more substtuton wll occur between domestc commodtes and mports and the sooner the mport markets wll be cleared. A hgher value for the foregn export demand elastcty leads to hgher welfare gans, because addtonal demand leads to ncreasng factor prces (compared to the foregn prce level) and therefore to ncreasng real ncome. The fact that trade elastctes are the most crucal elastctes, s a well known result n appled general equlbrum analyss (see Mensbrugghe et al., 1990, p. 200, for example). The remanng elastctes (elastcty of substtuton of the σ T, elastcty of substtuton of producton ( σ ), ) have households ( σ H ), transport elastcty ( ) vrtually no nfluence on natonal real ncome. P 16 We have performed senstvty analyss for flexble wages only. The results for fxed real wages are very smlar. The model results are calculated for fve elastcty values, whereby - one by one for each elastcty - base values are multpled wth the followng factors: 0.5, 0.75, 1, 1.5, and 2. 18

19 Fgure 2: Senstvty of real ncome n Austra to the varaton of elastctes for the counterfactual equlbrum 'CEEC Export Expanson' (per cent changes compared wth the benchmark equlbrum, varaton of elastctes compared to the base run value) 0.35% 0.30% 0.25% 0.20% 0.15% 0.10% 0.05% Senstvty to mport supply elastcty 0.00% % 0.40% 0.30% 0.20% 0.10% Senstvty to mport demand elastcty 0.00% % 0.30% 0.25% 0.20% 0.15% 0.10% 0.05% Senstvty to export demand elastcty 0.00% Senstvty to elastcty of substtuton of households 0.30% 0.30% Senstvty to transport elastcty 0.30% Senstvty to elastcty of substtuton of producton 0.25% 0.25% 0.25% 0.20% 0.20% 0.20% 0.15% 0.15% 0.15% 0.10% 0.10% 0.10% 0.05% 0.05% 0.05% 0.00% Source: The authors own calculatons. 0.00% % In addton the varaton of the elastctes stated above, we have nvestgated the nfluence of the transport rates on the regonal dstrbuton of the real ncome gans. Therefore we have ranked the Austran regons accordng to ther dstance to the CEEC-4. The regonal welfare results for the base run can be found n the second row n fgure 3. What we can obtan s that the regon wth the lowest dstance to the CEEC-4 (Venna) has the hghest welfare gans, the regon wth the hghest dstance (Vorarlberg) has the lowest welfare gans. In order to separate the nfluences of sectoral composton and dstance to CEEC-4 of Austra s regons on regonal welfare, we frst set the transport rate equal to zero. Ths solates the nfluence of sectoral composton (frst row n fgure 3). What we see s a great smlarty wth the results of the base run. Ths suggests that the sectoral composton of the regons manly determnes the regonal results. If we double the transport rate (thrd row n fgure 3), we see the growng nfluence of the transport rate. By assumng that all regons have the same sectoral composton, we solate the nfluence of dstance (fourth row n fgure 3). We see that Venna s welfare gans are 118 % of the natonal gans, Vorarlberg s gans are 72 % of the natonal gans. What we can conclude now s that both dstance and sectoral composton matters, but - wth our parameter values - the latter seems to have a greater nfluence. 19

20 Fgure 3: Regonal welfare gans for the Austran regons for the scenaro 'CEEC Export Expanson' for dfferent transport rates (percent of natonal values) 160% 140% 120% 100% 80% 60% 40% 20% 0% Venna Lower Austra Burgenland Styra Carntha Upper Austra Salzburg Trol Vorarlberg Same structure Transport rate *2 Base run Transport rate = 0 Legend: 'Transport rate = 0' No transport costs for nterregonal and nternatonal transport 'Base run' Base run of the model 'Transport rate*2' Double transport costs 'Same sectoral structure' The same sectoral structure s assumed for all Austran regons; ths shows the mpact of dstance to CEEC-4 Source: The authors own calculatons. 6. CONCLUSIONS In ths paper we have presented a framework to nvestgate the nfluence of economc development n Eastern Europe on regonal producton and welfare n Austra. We have concentrated on the effects future trade patterns between Austra and ts 4 neghborng Central and Eastern European countres (CEEC-4) wll have on welfare and producton for Austra's regons. Our framework s a multregonal general equlbrum model for the Austran economy. The 9 Austran Federal Provnces (NUTS-II) are lnked by trade flows wth 4 of the Central and Eastern European countres and wth the rest of the world. We have seen that the development of trade between the CEEC-4 and Western Europe after 1989 can be characterzed by three phases, namely the 'early restructurng phase' ( ), the 'recovery phase' ( /95) and the 'catchng-up phase' (1994/ /20). We have concentrated on one subset of possble trade developments durng the 'catchng-up phase' (1994/ /20). Therefore we have constructed the scenaro 'CEEC Export Expanson'. In ths scenaro we have assumed that the CEEC-4 wll ncrease ther supply of exports to Austra to the three-fold level of 1994 for all sectors. The represents a success of the CEEC-4 n buldng up new export capactes. We have examned welfare and producton effects of ths scenaro on Austra's regons for two dfferent model specfcatons (flexble and fxed real wages). Real ncome n Austra rses by 0.26% (flexble wages) and 0.86 % (fxed real wages). There s a 20