Industrial specialisation, trade, and labour market dynamics in a multisectoral model of technological progress. Robert Stehrer*)

Transcription

1 DEPARTMENT OF ECONOMICS JOHANNES KEPLER UNIVERSITY OF LINZ Industral specalsaton, trade, and labour market dynamcs n a multsectoral model of technologcal progress by Robert Stehrer*) Workng Paper No January 2001 Johannes Kepler Unversty of Lnz Department of Economcs Altenbergerstrasse 69 A-4040 Lnz - Auhof, Austra *) correspondng author: stehrer@wsr.ac.at phone +43 (0) , (fax)

2 Abstract The ssue of the mpact of trade on specalsaton structures and the effects of trade lberalsaton on employment and labour markets has been ntensvely dscussed n the recent lterature on trade lberalsaton and globalsaton. In Europe ths debate has ganed new momentum n the dscusson on the effects of the catchng-up processes of the transformng economes n Eastern European Countres. But the bulk of the exstng lterature n ths area employs almost wthout excepton a statc Heckscher-Ohln framework based on factor-endowment dfferences and thus seems not to be a sutable tool for analysng dynamc ssues of technology catchng-up and dynamc adjustment processes. In ths paper I present a model to explore the ssue of productvty catchng-up, nternatonal specalsaton and labour market effects n a dynamc mult-sectoral framework wth heterogenous labour. The model s bascally an nput-output model, but also has some Schumpeteran features. These Schumpeteran features are the mpact of transtory rents, emergng from (labour) productvty-enhancng technologcal progress or catchngup processes, upon the prce-, wage- and quantty system of the tradng economes. Relatve productvty and relatve wage rate dynamcs across sectors determne comparatve cost advantages and trade specalsaton. The second part of the paper presents some smulaton studes of the evoluton of prces, output, employment and wage structures, where varous stylzed types of technologcal progress and ndustral catchng-up processes are modelled. In the appendx of the paper the equlbrum solutons of the model are derved. JEL-Classfcaton: C62, C63, C67, D57, F15, F17 Keywords: trade lberalsaton, economc ntegraton, labour markets, smulaton, economc dynamcs, growth

3 Contents 1 Introducton 1 2 The model The basc N-sectoral model Technology Prces and rents Labour market Quanttes Extenson I: S skll-types of workers Extenson II: Tradng economes Exports and mports for consumpton Investment flows Quantty dynamcs Learnng processes Internatonal effects on prces The balance of payments Smulaton Studes Technologcal progress n a closed economy Assumptons Skll based technologcal progress n a closed economy Assumptons Smulaton results Effects of nternatonal trade and catchng-up The assumptons Smulaton results Conclusons 34 A Equlbrum, balanced and steady-state growth 37 A.1 Closed economy A.1.1 Technology A.1.2 Prces A.1.3 Profts and rents A.1.4 The quantty system A.1.5 Labour demand A.2 Integrated economes A.2.1 Prces, profts, and rents A.2.2 The quantty system References 45

4 Lst of Fgures 3.1 Labour nput coeffcents Prces and unt labour costs Rents Wages and consumer prce ndex Labour demand Output Labour nput coeffcents Prces and unt labour costs Rents Wage rates Labour demand Output Growth rate and consumer prce ndex Labour nput coeffcents Prces Consumer prce ndex Rents Wages Unemployment Labour demand Output Trade and nvestment flows Balance of payments Lst of Tables 3.1 Parameter values used n smulatons Startng values used n smulatons Startng values used n smulatons Parameter values used n smulatons Parameter values used n smulatons Startng values used n smulatons

5 INDUSTRIAL SPECIALISATION, TRADE, AND LABOUR MARKET DYNAMICS IN A MULTISECTORAL MODEL OF TECHNOLGICAL PROGRESS Robert Stehrer 1 1 Introducton Ths paper presents a dynamc multsectoral model to study the effects of technologcal progress, catchng-up and trade lberalsaton on the labour market performance of dfferent skll-types of workers n advanced and catchng-up economes. The ssue of the mpact of trade on labour markets n the more advanced economes was wdely dscussed at the begnnng of the 1990's, when a number of free trade agreements (especally the NAFTA between the US, Canada and Mexco) came nto beng. The debate focused manly on the mpact of developng countres and exporters of low-skll ntensve goods on the relatve wages of sklled to unsklled workers n the more advanced countres, especally the US. In ths debate the emprcal evdence of a large wdenng of the wage dfferental between sklled and unsklled workers n the US n the 1990's was the startng pont. On the one hand, especally Leamer (1994, 1996) and Wood (1995) argued that trade lberalsaton was the man reason for the worsened labour market poston of the unsklled workers. Ths explanaton was, on the other hand, crtcsed e.g. by Lawrence and Slaughter (1993) and especally Berman et al. (1994). The latter ponted to skllbased technologcal progress as the man explanaton for the labour market postons of low sklled workers. In the debate on the effects of trade on labour markets manly the statc framework of the Heckscher-Ohln model was used. In the case that the advanced country s relatvely better endowed wth sklled (relatve to unsklled) labour than the less-developed country, skll-ntensve goods are relatvely cheaper n the advanced country. In a free trade regme, ths country should then specalse n skll-ntensve goods, whch then rases the demand for the sklled workers, and thus relatve wages of sklled workers are ncreasng. The opposte s expected for the less-developed country, whch s relatvely better endowed wth unsklled workers. Theoretcal and emprcal studes then focused on the relatve mpact of trade versus technologcal progress. In most studes technologcal progress was found to have the most mportant mpact on the labour market performance of the lower-sklled versus the hgher sklled workers (measured ether n relatve wages or relatve unemployment rates). Usng factor content analyss, Wood (1995), Sachs and Shatz (1994), and others advocated for 1 Iacknowledge support from the Jubläumsfonds of the Austran Natonal Bank n the context of the project 'Technology, Productvty and Employment n the Accesson Countres'. I want to thank Mchael Landesmann and some of the partcpants of the IIOA conference n Macerata, Italy, August 2000, for useful comments. The author remans responsble for any errors that may reman. 1

6 the mportance of trade n explanng rsng wage dfferentals. Lawrence and Slaughter (1993) crtssed ths vew, as n the Heckscher-Ohln model relatve prces of goods must change, causng changes n relatve factor prces (Stolper-Samuelson effect). But they could not fnd emprcal evdence for such a change n relatve prces. On the technology sde, Berman et al. (1994) examned the mpact of technologcal progress on sklled relatve to unsklled workers. Fndng that relatve demand for sklled workers had rsen n each ndustry (ntrandustry versus nterndustry shfts n relatve demand), they concluded that skll-based technologcal progress was much more mportant than the effects of trade lberalsaton. Further Feenstra and Hanson (1996) showed that relatve wages of the sklled workers n Mexco (the country relatvely better endowed wth unsklled workers) has also rsen (contrary to the expected effect of the Heckscher-Ohln model). In explanng ths fact, they used a Heckscher-Ohln model, where outsourcng actvtes may lead to ncreases n relatve wages of the sklled workers n both countres. From the vewpont of ths paper there are several drawbacks n analysng the ssue of trade and labour markets wth the models mentoned above. The man crtcsm of the Heckscher-Ohln framework can be summarzed n two tems: Frst, ts genune statc nature and, second, the assumpton of equal technologes n the countres. 2 In ths paper we shall start from the evdence that countres use dfferent technologes (expressed here as labour productvty levels) but are able to catch-up to the more advanced countres. The perod from the start of the catchng-up process to reachng the technology fronter s by tself nterestng and worth studyng, but the shape of the transton and the postonng wthn ths perod can have long-term effects. Landesmann and Stehrer (2000b) fnd that the half-tme of catchng-up n labour productvty levels n dfferent manufacturng sectors ranged from years, a tme perod whch should not be neglected n analytcal research. Further the catchng-up process s dfferent across sectors. From ths vewpont a model where comparatve dynamc analyss can be made, rather than purely comparatve statc analyss, seems to be justfed. Further, the Heckscher-Ohln framework reles only on dfferences n factor endowments whle assumng equal acces to the nternatonal technologcal standards. Although ths assumpton seems to be relevant (at least partly) for trade between advanced and developng countres, t seems less satsfactory when studyng effects of trade lberalsaton between Eastern European countres and the EU, as here the dfferences n factor endowments seem not be that large (or even reversed, f one compares wth EU-Southern countres only). In the model presented below, a Rcardan framework wth catchng-up n labour productvty levels s used, where also dfferences of payments to factors of producton can be ntroduced. In Europe ths ssue s now debated wth respect to the ntegraton of the Eastern European countres and the mpact on the labour markets of the EU and the Eastern European countres. Although there are also large dfferences n (labour) productvty, some countres manage to catch-up qute rapdly to the level of the Western European 2 Ths means that the countres are on the same soquant, although they are on dfferent ponts on ths soquant due to dfferent factor endowments and dfferences n relatve factor prces n the autarkc equlbrum. 2

7 countres, whereas some others are stayng or even fallng behnd. In two recent papers (Stehrer and Landesmann, 1999; Landesmann and Stehrer, 2000b) we focused on the technologcal performance of a rather large sample of catchng-up economes (European Southern countres, Eastern Asan countres, etc.) at a dsaggregated manaufacturng level. The man fndngs were that, frst, there are huge dfferences n labour productvty levels between the developng and the developed countres, and, second, these dfferences are closed qute rapdly by catchng-up processes (at least for some countres or country groups). Further the relatve speed of catchng-up n levels of labour productvty was found to be qute dfferent f one compares dfferent manufacturng (e.g. hgh-to-medum tech and lower-tech) sectors. Ths paper presents a dynamc, multsectoral model of catchng-up, where these ssues can be dscussed n an ntegrated framework. Further the model allows to dscuss labour market effects on dfferent skll-types of workers. The man focus of ths model s to analyse the mpact of 'shocks' (ether technology or trade lberalsaton) and t thus deals manly wth non-steady-state and non-balanced growth and fluctuatons. Thus t s not the am of ths paper to study long-term steady-state dynamcs, rather we study traverses from one long-term equlbrum to another long-term equlbrum. Further some lmtatons of ths approach but also further research ssues are dscussed at the end of the paper. The paper goes as follows: In the frst part, the dynamc model used n the smulaton studes s presented. Ths s done step by step: Frst we ntroduce the basc model wth one autarkc country and homogeneous labour. Then the model s extended to the case wth heterogeneous labour and fnally to a trade model wth nteractng countres. In the appendx the long-term dynamc equlbrum propertes of the model are dscussed. Ths s useful as the 'behavour' of the model and some specfc assumptons n the nonequlbrum transton phase become clearer, f the long-term propertes are accounted for. In the second part of the paper three smulaton studes are presented. The frst one shows the mpact of sector specfc technologcal progress n a closed economy wth homogenous labour. Ths also helps to nterpret the dynamcs of the system n more complcated envronments (e.g. wth more skll-types of workers, more countres, etc). The second smulaton run then shows the model wth two types of labour (sklled and lower-sklled workers). Fnally, the thrd smulaton dscusses a model wth two tradng economes, where one of them catches up n terms of productvty levels wth the more advanced economy. Here especally the mpacts on the labour markets of both countres are dscussed. The contrbuton of ths paper s manly to ntroduce ths knd of dynamc structural modelng n a very smple way,.e. usng qute smple specfcatons whch can be mproved n further research. Extensons of the model ncludng senstvty analyss and applcatons to specfc topcs whch are to be analysed n future research. 3

8 2 The model 2.1 The basc N-sectoral model In ths secton we present the detaled structure of the model, whch s used afterwards n smulaton studes. To be more explct on the equatons n ths secton we do not use matrx notaton. The model s based on a paper by Landesmann and Stehrer (2000a) and s an extenson and modfcaton of the model presented theren n a number of respects Technology We start wth a smple matrx of techncal nput coeffcents, denoted by A = 0 a 11 ::: a N a 1N ::: a NN whch s assumed to be stable over tme. Labour productvty s gven by a vector of labour nput coeffcents a l =(a l1 ;:::;a ln ) Labour s used n fxed proportons. But these labour nput coeffcents a l may declne over tme at an exogenous rate g al» 0toa predetermned level μa l : 1 C A _a l = g al (a l μa l ) (2.1) Ths formulaton mples that a l > 0 for all t and thus, that labour s seen as a necessary nput for the producton of each good. Further the growth rate of labour productvty s gong to 0, _a l a l! 0 when t! Prces and rents Prces where Prces are modeled as adjustment to unt costs _p = ff p [(1 + ß)c p ] (2.2) c = X j p j a j +! are the costs of producton and! = w a l denote the unt labour costs (for the moment we lmt the analyss to one skll-type of workers). We assume that wage rates w need not be equalsed across sectors, although 3 The growth of labour productvty could also be made endogenous. In ths paper, however, we emphasze the effects of technologcal progress rather than the sources of productvty growth. 4

9 t s possble n general that wage rates equalse n the long run as we shall see below. The parameter 0» ff p» 1 gves the speed of adjustment of prces to (equlbrum) unt labour costs. There exsts a long run mark up on prces ß whch s equal across all sectors (ths leads to equal real per unt profts n each sector). Rents As there s a constant mark-up on prces ß there are long run per unt profts r defned as r = ßc As prces do not adjust mmedately to unt costs plus mark-up there arse transtory profts dependng on the speed of technologcal progress g al and the prce adjustment parameter ff p and the dynamcs of wages as we shall see below: s = p (1 + ß)c = p c ßc = p c r Prce ndex For consumers the prce ndex s an mportant ndcator. The consumer prce ndex s defned as P C = X ff p wth X ff =1.e. a weghted sum of prces (weghts are the nomnal shares of consumpton ff whch are ntroduced more specfcally below) Labour market Wage rates Nomnal wages are also growng or fallng as (especally transtory) rents are partly dstrbuted to workers and because of excess supply (demand) of workers n the labour market: _w =» s s a l +» u uw +» w w μw w (2.3) 0»» s» 1 s the proporton of per unt (transtory) rents s pad to workers. Rents are dstrbuted only to workers n the respectve sector where the rents arse. (A more general formulaton would also allow that wages of workers n other sectors to rse due to profts n a partcular sector.) The second term on the rhs of the wage dynamcs equaton reflects the mpact of unemployment on the dynamcs of the wages (» u» 0), where unemployment s defned as u = LS P a lq L S = L S L D L S Thrd, there s an mpact on the wage dynamcs f wages (for the same type of worker) dffer across sectors. Ths reflects the common assumpton (e.g. n the standard Rcardan trade model), that wages get equalsed across sectors because of labour moblty. The (weghted) average wage μw s defned as μw = P P LD w 5 LD

10 If the average wage μw s hgher than the sectoral wage w the wage n sector wll rse (_w > 0 for» w < 0), n the other case fall. Ths term works across all sectors. Thus n the formulaton used n the smulatons, there s a sector specfc term and two economy wde terms havng an nfluence on wage rates n each sector. There can occur wage dfferentaton across sectors n the short run; wages are equalsed, however, n the long run. Labour supply Labour supply L S s assumed to adjust to labour demand accordng to _L S = ff L S L D L S (2.4) where L D = P a lq and ff L S =0 for L S >L D ff L S 0 for L S» L D Ths formulaton mples that labour supply s adjustng to labour demand f there s excess demand of labour, but there s no adjustment n the other drecton;.e. that workers leave the labour market n case of unemployment (excess supply of labour). 4 In the smulatons below t was actually assumed, that there can be no excess demand for labour as labour s suppled wth nfnte elastcty and adjusts mmedately to labour demand. 5 Ths assumpton can be justfed for two reasons: Frst, there s some evdence that shortage of labour has been not acted as a constrant n the long run growth of economes (e.g. McCombe and Thrlwall, 1994). Further wth applcaton to actual catchng-up processes of some countres a shortage of labour was never dscussed as lmtng factor, as ether the labour supply responds suffcently fast to the growth process or labour s avalable from other sectors (as for example n the model by Lews, 1954). Second, from a modellng pont of vew, a constrant of labour supply would mply a further specfc assumpton on the dstrbuton of labour across the sectors, whch s especally a dffcult problem when assumng more than one skll-type of workers Quanttes After ths dscusson of the prce system the quantty system must be specfed. Demand for goods conssts of three dfferent components whch can be summarzed n the followng 4 A less strong assumpton would be that the parameters dffer for the two stuatons, so that hgh unemployment leads to a fallng partcpaton rate. 5 Thus the producton s not constraned by shortages n the supply of labour, although n the smulatons we allow for a pressure on wage rates due to excess demand of labour va the unemployment term. Ths excess demand of labour results from the numercal soluton of the system of dfferental equatons. As labour supply adjusts rapdly to demand (ff L S = 1 for L S» L D ) ths effect may not be very large. 6 The lmtatons of (sectoral) growth due to a shortage of factors would of course be an nterestng topc n tself, but s not a topc n ths paper. Thus, research n ths area has to be postponed. 6

11 general equaton: q D = X j = X j X (1» sj )s j + r j X a j q j + f q j + ff a j q j + q I + q C j p j w j p a lj q j (2.5) wth P ff = P f =1. The frst term s demand for ntermedate goods used n producton, the second term s demand out of profts (whch are entrely used for nvestment) and the thrd term reflects demand out of workers ncome (used for consumpton). (The terms q I and q C wll be dscussed below n detal). Thus q I + q D s the fnal demand for good. Specfcally we assume further that workers do not save ther ncome (or spend all money on consumpton goods), whereas profts are entrely used for nvestment. 7 Further t s assumed that nvestments cannot be negatve (see below). Consumpton demand For consumer P demand we assume that the nomnal shares of consumpton ff are constant and ffq =1. Or, stated dfferently, consumers maxmze a Cobb-Douglas utlty functon, U = qff, from whch ths knd of consumer behavour results. Of course, each other demand system - e.g. Stone-Geary, CES utlty functons, AIDS demand system, or Dxt-Stgltz type - whch gves nomnal shares for gven ncome and prces - and thus would allow non-lnear Engel curves, varous prce elastctes, and so on - could be used nstead of the smple Cobb-Douglas system. 8 Investment demand Investment demand s smlarly formulated wth nomnal shares f whch allocate nvestments to the dfferent sectors. Ths specfcaton descrbes only the aggregate outcome of nvestment decsons at the frm or ndustry level whch s not explctely formulated. As the structure of the economy changes over tme (due to changes n relatve prces, real ncomes, and thus consumpton patterns 9 the structure of nvestments also has to change. In ths paper we shall assume that the nvestment structure adjusts to the the growth maxmsng structure f Λ over tme. The specfc formulaton for the nomnal shares of nvestment n the smulaton model s _f = ff f Λ (f f Λ ) (2.6) where f Λ = p q P p q Ths formulaton assumes myopc decson behavour. For gven quanttes and prces at tme t one can calculate the growth maxmsng nvestment structure, f Λ. The actual 7 Ths assumpton s not necessary but smplfes dscusson and presentaton of the model. E.g. f part of the profts s used for consumpton wth dfferent structure as workers consumpton, another term would have tobentroduced n the equaton above. 8 A specfcaton of a utlty functon would also allow for analyss of welfare mplcatons. 9 Although when usng the smple Cobb-Douglas demand changes n real ncome do not affect the structure of consumpton. 7

12 nvestment structure then adjusts to ths optmal one gradually. 10 sector are then gven by q I = max ψ 0;f X j! (1» sj )s j + r j q j p Investments n one (2.7) where we assume that nvestment must be non-negatve. The growth rate of the economy depends on ths nvestments as we dscuss n the next step. Supply of goods The supply of goods s then modeled as an adjustment process where supply adjusts to demand n a growng economy wth: " X # _q = ff q (1 + g) ~a j q I j + q C j q (2.8) j s an adjust- where ~a j denotes atypcal element of the Leontef nverse [I A] 1 and ff q ment parameter. The overall growth rate of the economy s determned as g = mn q I q The ratonale for ths specfcaton s as follows: At each pont n tme there exsts a fnal demand vector, q I + q C. For the system to be able to produce these quanttes the ntermedate demand for the producton of each good must be taken nto account, whch s done by the Leontef nverse. Further the economy s able to grow only f there are postve nvestments n each sector, whch amount - n ths model wth crculatng captal only - to a growng stock of ntermedate nputs. Due to the lnearty ofthe producton system the overall growth rate s bounded by the sector wth the lowest nvestment rate (.e. the rato of qi q ). In the Appendx A we show that n equlbrum (.e. steady state balanced growth) the nvestment structure wth nomnal shares f Λ as defned above guarantees that the system grows wth g Λ = ß. 1+ß 2.2 Extenson I: S skll-types of workers A smple but very nterestng extenson of the basc model above s the assumpton that there are more than one type of worker (e.g. hgh- and low-sklled workers). In fact the model can n ths respect be generalsed qute easly. Startng wth the prce equaton one 10 In dscrete tme one could calculate an optmal path by an teratve process: Gven prces p t and quanttes q (1) t one calculates n the frst step (1) the parameters f (1). Insertng these nto the quantty system yelds new quantty values q (2) t, whch agan leads to new parameters f (2), and so on. Ths teratve process at each pont n (dscrete) tme then leads to the optmal nvestment structure f Λ whch maxmses the growth rate (see below and the dscusson n Appendx A). 8

13 only has to change the expresson for unt labour costs. The ndustry specfc unt labour costs are now the sum of the products of the skll specfc wages and nput coeffcents:! = X z w z az l (2.9) where z denotes the skll types of workers, z = 1;:::;S. If wages by skll groups are equalsed across sectors the unt labour costs by sectors are! = X z w z a z l and thus the dfferences n sectoral unt labour costs depends only on dfferences n the productvty levels and the structure of labour nputs. Further the technologcal progress can be dfferent across skll groups, thus the dfferental equaton for the labour nput coeffcents s _a z l = g a z l (a z l μa z l) The effects of transtory rents and unemployment on wage rates are now becomng skllspecfc whch s one of the most nterestng characterstc of the extenson of the model to more than one skll group. As the unemployment rate s dfferent across skll groups and has equal mpact on wage rates n the varous sectors the wage rate dynamcs dffers not only from sector specfc rents but also because of dfferent skll-ntensty and dfferng unemployment rates. In the smulatons below we assume the followng specfcaton: where _w z s =» s z P +» u zu z w z w z z az +» w μw z z l w z P LDz w z w z wth» s z =» s P z wz (2.10) μw z = P LDz The specfcaton of the frst term on the rhs of equaton (2.10) mples that wage rates of dfferent skll types of workers wthn an ndustry are growng at equal percentage rates. Ths means that wage rates can (temporarlly) be dfferent across sectors and skllgroups. The second term gves the mpact of the skll-specfc unemployment rates u z on sectoral wage rates by skll group. The thrd term agan rests on the assumpton that wage rates for the same skll-type of workers equalse n the long run as dscussed above. The unemployment rate u z has to be defned skll specfc too: u z = LS z P az l q L Sz = L S z L Dz L Sz Agan we assume as above that labour supply adjusts to demand accordng to where L Dz = P ;z az l q and _L Sz = ff L S z L D z L Sz (2.11) ff L S z =0 for L Sz >L Dz ff L S z 0 for L Sz» L Dz 9

14 The reasons for ths assumpton are the same as above. But here one has to notce that the assumpton s even more restrctve for sklled workers (whch have to be educated) than for low- or unsklled workers. Fnally the demand component ofworkers has to be changed n the demand equatons to XX wj z ff a z lj p q j j z Ths specfcaton assumes that the nomnal consumpton shares are equal for all skll groups. In an even more general settng the nomnal shares may depend on skll- and sector-specfc wages and prces (e.g. when usng demand systems wth non-lnear Engel curves). 2.3 Extenson II: Tradng economes The next step s to ntroduce more countres and especally nternatonal relatonshps between these countres. Frst of all, all the varables have to be ndexed for the dfferent countres. In ths paper we restrct the analyss to a two country model and denote the varables wth L for the leader and r for the other country, respectvely. 11 If there are general relatonshps between the two countres we denote them by r and s, respectvely. Further we assume that the exchange rate between the tradng economes s set to X rs =1 and ther are no changes over tme. (Effects of trade mbalances on exchange rates shall be ntroduced at a later stage.) Then the varous economc relatonshps between the countres have to be specfed. Three dfferent ways of economc lnkages are specfed n ths paper: mports and exports, nvestment flows, and nternatonal learnng processes Exports and mports for consumpton For consumer demand we adopt a specfcaton whch s smlar to the specfcaton n the closed economy P case. Consumpton demand n country r now depends also on ncome n country s, Pz wz;s L Dz;s. For smplctywe assume that a constant nomnal share of wage ncome μ sr n country s s spent on goods from country r. The nomnal share of ncome n economy s spent on goods n economy r can then be wrtten as ff sr = ff s μsr P where r P ffsr =1must be satsfed. 12 Ths specfc assumpton means that the domestc and foregn good are not (or not seen as) perfect substtutes. In fact, the formulaton 11 See appendx A for a dscusson of the general case. 12 In the two country case ths means that ff ss =1 ff sr 10

15 used here mples a Cobb-Douglas utlty functon of the form 13 U s = Y ;r (q r ) ffsr where the prce elastcty equals Investment flows Investors have to make two decsons: Frst, n whch country and sector to nvest, and second where to buy these nvestment goods. The decsons for these two questons are guded by dfferent questons: The frst one s motvated, where the hghest per unt rents (and profts) can be gathered, the second where the goods for nvestment can be bought relatvely cheaply. These goods have then to be transported to the country where they should be nvested. (In ths case also transport costs may be consdered or neglected by assumpton as above for consumpton goods.) But, for smplcty, we assume n ths paper, that f an nvestor wants to nvest n a certan country, the goods for nvestment are also bought n that country (the money s spent abroad). Thus the sum of rents and profts n a country s n sector, (1» s s )s s + rλ s q s, s then dstrbuted accordngly to demand from rents and a smlar specfcaton as for consumpton demand can be used. We assume that the nomnal share spent abroad s f sr = f r νsr P where ;r fsr = 1denotes the nomnal share of nvestment expendture of country s n country r n sector. For smplcty we assume agan that the nomnal share of rents and profts whch are spent abroad s constant ν sr. 14 The dstrbuton of nvestment expendtures across sectors s modeled as before: _f r = ff r f Λ (fr f rλ ) where f rλ = pr qr P pr qr Note the dfference between the nomnal shares for consumpton and the nomnal shares for nvestment expendtures. In the former case the nomnal share of consumpton of country s n country r, ff sr = ff s μsr, depends on the consumpton structure prevalng n country s, ff s, whereas the nomnal share of nvestment of country s n country r, 13 Of course here agan more flexble functons could be used whch e.g. allow for home-bas effects, other prce elastctes between foregn and domestc goods, etc. Further one could also ntroduce a specfcaton that these nomnal shares evolve gradually n the case of a sudden trade lberalsaton. 14 In a more advanced specfcaton ths share may be dependent e.g. on (sector specfc) rents across countres. But there are a number of motvatons for foregn drect nvestments that the chosen partcular formulaton may be a good frst approxmaton. Further ths formulaton also mples foregn drect nvestment from the less developed country n the advanced country whch could occur e.g. n case of repatraton of profts. 11

16 f sr = f r νsr, depends on the nvestment structure n country r, f r. From ths follows that the output structure of an economy s also nfluenced by the consumpton structure n another economy (f the nomnal shares are dfferent). On the other hand, the structure of output does not depend on the decsons n the economy abroad; only the overall growth rates are nfluenced by the nvestment flows across economes. Further we assume that ntermedate nvestments are not traded Quantty dynamcs Gven ths assumptons on the consumpton and nvestment behavour n the nternatonal settng the demand for products n country r can then be wrtten as: X q D r = j a r j qr j + X s X j f sr (1» s s j )s s j + r s X j p r q s j + s XX where agan the nvestment n a partcular sector has to be constraned wth q I r = max ψ 0;f r XX and the growth rate of a partcular economy s then s g r = mn j ψ q I (1» s s j )s s j + r s j q r r! p r j z q s j ff sr! w z;s j p r a z;s lj qs j (2.12) The supply adjusts to demand dfferental equaton s then " X _q r = ff r q (1 + g r ) ~a r j j q I r r # j + q C j (2.13) Learnng processes The nternatonal lnkage of economc ntegraton s that countres can learn from each other, meanng that technologcally backward countres are catchng-up wth more advanced countres. For ths dfferent approaches could be regarded. The smplest modellng strategy, whch s used n ths paper, s that countres are catchng-up to the leadng country (or the productvty fronter). Dfferent paths of catchng-up processes were nvestgated n Landesmann and Stehrer (2000b) and should not be repeated here. In the smulatons above we assume that countres lyng farther behnd have relatvely hgher productvty growth rates (Gershenkron's 'advantage of backwardness' whch s appled here at the ndustral level; see also Landesmann and Stehrer (2000b) for a theoretcal dscusson and emprcal analyss). The specfc equatons for the catchng-up processes are smlar to the closed economy case: _a r l = g r a l a r l μa L l 12 (2.14)

17 where μa L l denotes the labour nput coeffcent of the technologcal fronter of the productvty leader. In a more sophstcated settng, the speed of catchng-up could also depend on the country-wde or ndustry-specfc skll-structure (relatve to other countres), exogenously gven learnng parameters, the structure and volume of mports and exports and especally the flows of nternatonal nvestments Internatonal effects on prces The last effect of nternatonal trade s that goods prces p r may equalse n the long run ('law of one prce'). In the set-up of the model so far a long-term equlbrum could exst wth persstent dfferences n prces, as the producton structure may change to the equlbrum structure n each country and there s no effect on prces va excess supply or demand. In the followng we therefore assume an exogenous trend for prce equalsaton. Ths alters the system of dfferental equatons for prces whch becomes now where _p r = ff r p [(1 + ß)c r p r ]+ff r μp p r μp p r μp = P P r qr pr r qr (2.15) saweghted average of the prces n the world market. 15 Further the consumer prce ndex must be defned for the nternatonal case. Gven the assumptons on the consumpton structure P C;r s gven by: The balance of payments P C r = X r X ff rs ps The goods demanded from country r n country s n a partcular sector are m rs = ff rs XX j z w z;r j p s a z;r lj qr j whch s mport demand of country r from country s. The exports of country r to country s are then denoted by x rs (= m sr ) The trade balance of country r wth country s, b rs, s then defned as the value of exports mnus the value of mports of country r: b rs = X (x rs pr m rs ps ) 15 In a more advanced settng the formulaton of the long-term prce equalsaton could be dependent on the mport shares n each country and sector, respectvely. 13

18 and thus denotes the value of net exports. The quanttes demanded for nvestment from country r n country s are n rs X = f rs j (1» s r j )s r j + r r j p s q r j whch denotes a 'real' FDI flow from country r n country s. country s to country r are then denoted by Investment flows from y rs (= n sr ) Smlarly one can specfy the value flows for nvestments as k rs = X (y rs p r n rs ps ) The balance of payments (ncludng trade and captal flows) s then the sum of the trade and the captal account BoP = b rs + k rs 3 Smulaton Studes In ths secton we present some prelmnary smulaton studes. 16 Frst, we present a smulaton for a closed economy and analyse the effects of sector based techncal progress n a model wth homogenous labour. The second set of smulatons then analyses skllbased techncal progress n a closed economy. In the thrd part, we ntroduce trade wth a second country and show smulatons of trade and nternatonal technology spllovers. 3.1 Technologcal progress n a closed economy Assumptons The frst smulaton shows the effect of exogenous technologcal progress n a closed economy. For smplcty we assume the followng partcular parameters. The technology matrx s a11 a A = 12 = a 21 a 22 0:40 0:10 0:10 0:40 Ths means that the technology matrx s symmetrc 17 and half of the output of perod t must be renvested to ensure the reproducton of the system at the same level. Labour productvty s equal n both sectors a =1 16 The model was wrtten and smulated n DMC (Medo, 1992). A Runga-Kutta algorthm s used for numercal smulatons. 17 Ths rather specfc assumpton facltates the nterpretaton of the dynamcs below but does not restrct the generalty of the model. 14

19 Further wages are also equal and set to w =1 Gven that the mark-up 18 s ß =0:0 ths gves equlbrum prces p =2 The mplcaton of ths assumpton s that the economy does not grow over tme, as the economy s on the maxmum level of consumpton and thus the economy exactly reproduces tself. Investment and thus output growth n the economy only occurs f there s a devaton of unt costs from prces, meanng that real wages are below ther maxmum value. In the smulaton below, growth can thus only occur due to (exogenous) technologcal progress (modeled as reductons n labour nput coeffcents). Further we assume that the nomnal shares for nvestment and consumpton are ff =0:5 and f =0:5. The long run equlbrum output structure s then q 1 q 2 = 1. I assume that the startng values are q = 1. In equlbrum thus 0:5q has to be renvested for the reproducton of the system. The wage demand s - gven the parameters and prces - 0:5q n each sector. The parameters and startng values used n the smulaton below are summarsed n Tables 3.1 and 3.2, respectvely. The smulaton run we present below starts from Parameter Values Sector specfc Sector 1 Sector 2 a a j ß ff p ff q » s ff f Λ ff μa l g al Economy wde ff L S (L S» L D ) ff L S (L S >L D ) 0.000» u » w Table 3.1: Parameter values used n smulatons ths long-term equlbrum, but n sector 1 occurs exogenous technologcal progress. As mentoned above ths s mplemented as _a l1 = g al1 (a l1 0:5) Thus the labour nput coeffcents dmnshes to a level of μa l1 = 0:5 wth a decreasng growth rate, as _a l1! 0 for a l1! μa l1. Fgure 3.1 shows the tme trajectores of the labour 15

20 Varable Values Sector specfc Sector 1 Sector 2 a l w L D ! p c r s q I q C q Economy wde L D L S u g P C Table 3.2: Startng values used n smulatons AL1 AL Tme Fgure 3.1: Labour nput coeffcents 16

21 nput coeffcents. The labour nput coeffcent s gradually fallng n sector 1 to a level of a l1 =0:5 but s stayng constant n sector 2. Thus techncal progress s sectorally based. The fallng labour nput coeffcent a l1 has an mpact on labour unt costs n ndustry 1 and thus an effect on costs and prces n sector 1, and va the nput-output matrx A also on sector 2. The mpact of wage rate movements on costs and prces are dscussed below. Fgure 3.2 presents the resultng tme trajectores for prces and unt labour costs. Unt P1 P2 ULC1 ULC Tme Tme Fgure 3.2: Prces and unt labour costs labour costs n sector 1 are fallng to a level of about 0.5. In sector 2 unt labour costs are fallng less and even rsng n the later perods. Ths rse s manly due to the growth of wage rates n sector 1 and the equalsaton of wages across sectors. Ths dynamc results n changes n relatve prces. Prce p 1 s fallng much faster and to alower level than the prce n sector 2, p 2, so that the goods of sector 1 are becomng relatvely cheaper. But, as mentoned above, prces do not adjust mmedately to unt costs. In the transton, postve unt rents arse n sector 1 whch are partly dstrbuted to wages, (» s > 0), and, on the other hand, are renvested accordng to equatons (2.6) and (2.7) gven above. The evoluton of the rents can be seen n Fgure 3.3. In sector 1 rents are rsng rapdly due to the effects of technologcal progress and sluggsh adjustment of prces and wage rates, whch are dscussed below. In sector 2 transtory rents become negatve. The reason for ths s that wages n sector 2 are rsng because of the wage equalsaton across sectors (» w < 0). Ths rases wage rates n sector 2 and thus unt labour costs, whch together wth sluggsh prce adjustment leads to negatve transtory rents. 19 The developments of rents have an mpact on sectoral wages (together wth the unemployment rate and wage equalsaton). Further, the consumer prce ndex P C together wth the nomnal wage rates are mportant ndcators for welfare mprovements. The evoluton of wage rates and the consumer prce ndex can be seen n Fgure 3.4. Wage rates 18 Ths specal assumpton on the value of the long-term mark-up should be nterpreted as modelng the evoluton of the economy along a long-term growth path and the devatons from t n the case of exogenous shocks (technologcal progress or trade lberalsaton). 19 Of course, wth mark-up prcng such negatve transtory rents would mean that total rents are less than the long-term average. But the producers n sector 2 are gong to suffer further from substtuton effects on the demand sde. 17

22 S1 S Tme Fgure 3.3: Rents W1 W2 CPI Tme Tme Fgure 3.4: Wages and consumer prce ndex 18

23 n sector 1 are ncreasng due to the relatvely hgh transtory rents. Wage rates n sector 2 are also ncreasng because of a tendency towards wage equalsaton across sectors, but the ncrase s less fast (as there are no or even slghtly negatve transtory rents). Fnally, the unemployment rate has an mpact on wage rates n both sectors. As wage rates n both sectors are affected symmetrcally by the unemployment rate ths does not have an mpact on relatve wages, but delays adjustment of wage rates, keepng the transtory rents hgher and thus rasng nvestment and the overall growth rate of the economy. The dynamcs of sectoral and aggregate labour demand s shown n Fgure 3.5. Labour LD1 LD Tme Fgure 3.5: Labour demand demand n sector 1 s slghtly fallng n the frst phase, but then starts rsng due to the overall growth of the economy and the fallng speed of technologcal progress. Labour demand n sector 2 s rsng. Ths results n a hgher level of overall employment and an ncrease n relatve employment n sector 2. The dynamcs of labour demand results from the dynamcs of labour nput coeffcents and output. On the other hand, the dynamcs of output and the output structure s determned by consumer behavour and nvestors behavour: Consumers demand relatvely more of good 1, whch has become relatvely cheaper (substtuton effect). The evoluton of output and the overall growth rate g can be seen n Fgure 3.6. Output s rsng faster and acheves a hgher level n sector 1 n the new equlbrum than output n sector 2. The reason for ths s the substtuton effect n the demand for the goods: as goods of sector 1 are becomng relatvely cheaper, demand s rsng relatvely more and thus output grows faster. In ths smulaton, the effect of the fallng labour nput coeffcents on labour demand s stronger than the effect of output growth and demand substtuton n the ntal phase leadng to a temporary job-loss n sector 1 (the substtuton effect have n the current verso of the model been lmted to the fnal demand whle ntermedate demand nput coeffecents are held constant; substtuton effects could also be extended to ths area of demand). 19

24 Q1 Q2 G Tme Tme Fgure 3.6: Output 3.2 Skll based technologcal progress n a closed economy Assumptons The second smulaton captures the effects of skll labour-based techncal progress, whch s seen to be the man source of the worsened unemployment poston of unsklled workers n the recent debate. 20 The smulaton s started at the values gven n 3.3. As one can Varable Values Sector and skll specfc Sector 1 Sector 2 Sklled Unsklled Sklled Unsklled a z l w z L Dz Sector specfc Sector 1 Sector 2! p c r s q I q C q Economy wde, skll specfc Sklled Unsklled L Dz L Sz u z Economy wde g P C Table 3.3: Startng values used n smulatons see from the startng values of the labour nput coeffcents, sector 1 s the skll-ntensve 20 It would of course be nterestng to study the effects of skll-neutral but sector-based technologcal progress. But due to lmtatons of space, ths comparson shall be postponed to later research. 20

25 sector. As wage rates for sklled workers are relatvely hgher then the wage rates for unsklled workers, ths leads to a relatve hgher prce of good 1 and - gven the current Cobb-Douglas specfcaton of the demand sde - less demand relatve to good 2. As the technology matrx s symmetrc, the relatve demand for good 1 s smaller than 1, due the substtuton effect n the formulaton of consumers demand. The parameter values are gven n Table 3.4. The most mportant assumpton here s that the technologcal Parameter Values Sector specfc Sector 1 Sector 2 a a j ß ff p ff q » s ff f Λ ff μa z l g a z l Sector and skll specfc Sector 1 Sector 2 Sklled Unsklled Sklled Unsklled Economy wde, skll specfc Sklled Unsklled ff LS z (L Sz» L Dz ) ff LS z (L Sz >L Dz ) » u z » w z Table 3.4: Parameter values used n smulatons progress s based aganst the unsklled workers and dentcal for both sectors. The nput coeffcent a l, =1; 2, s fallng from 2.5 to a level of 2 wth a rate of g al = 0:015. The labour nput for the sklled workers are constant. Ths specal assumpton means that technologcal progress affects both sectors n the same way. Dfferences thus arses only from dfferences n the relatve skll-ntensty and the wage dynamcs n both sectors Smulaton results We shall now dscuss the most nterestng features of ths factor based techncal progress. Fgure 3.7 shows the trajectores of the labour nput coeffcents. These are fallng n both sectors to a level of a u l =2:000. The resultng dynamcs of prces and relatve prces can be seen n Fgure 3.8. Unt labour costs and prces are fallng n both sector. But the relatve prce of good 1 (the skll ntensve good), p 1 =p 2 s rsng for two reasons: Frst, the share of unsklled workers n total unt labour costs s hgher n sector 2 and thus unsklled labourbased techncal progress has a larger mpact n ths sector. Second, wages of unsklled workers are under more pressure, as there s unemployment for unsklled workers, but not for sklled ones, as we shall see below. Ths leads to a wage dynamcs whch s also n favour of the sklled workers, thus enhancng the effects on relatve unt labour costs. Rents (Fgure 3.9 are rsng and then fallng gradually to a zero level due to the fallng 21

26 AL1S AL1U AL2S AL2U Tme Tme Fgure 3.7: Labour nput coeffcents P1 P2 P1/P Tme Tme Fgure 3.8: Prces and unt labour costs S1 S Fgure 3.9: Rents 22

27 labour nput coeffcents and sluggsh adjustment of prces and wage rates. Rents n sector 2, whch s the unsklled-ntensve sector, are beng hgher than n sector 1. The outcome on the labour market, whch s now dvded nto sklled and unsklled workers, are shown n Fgures 3.10 and 3.11, whch present the trajectores for wages and labour demand. Wage rates for sklled workers (Panel A n Fgure 3.10) are ncreasng n W1S W2S W1U W2U Fgure 3.10: Wage rates both sectors, frst because there are postve rents, whch are partly dstrbuted to workers, and second, labour demand s rsng for sklled workers (as wll be dscussed below). The shortage of sklled labourers have a slghtly postve mpact on the wage rates for sklled workers. The labour market stuaton for unsklled workers s worsenng. As one can see (Panel B n Fgure 3.10) the wage rates of unsklled workers are fallng to a lower level. Although there are postve rents, whch are also dstrbuted to the unsklled workers, ths effect s - gven the parameter values - smaller than the effect of the unemployment rate, whch arses due the unsklled labour-based techncal progress. Fgure 3.11 shows the tme trajectores for employment levels n both sectors and for both types of workers, the skll-specfc unemployment rates and the relatve labour demand. Frst, labour demand for the sklled workers s rsng n both sectors. As we shall see below the growth rate of the economy becomes postve and thus rases the demand for sklled workers, together wth the fact that for ths group does not occur any labour-savng techncal progress. Labour demand for sklled workers s rsng relatvely more n sector 2 than n sector 1, as demand shfts to good 2. Further, demand for unsklled workers s frst fallng (the effect of the techncal progress s stronger the effect of the overall growth of the economy), but then starts rsng, as the technologcal progress s becomng slower and thus the overall growth rate of the economy leads to a postve growth rate of the demand for unsklled workers. Further the overall growth of the economy due to nvestments s too low, to boost demand for unsklled workers enough. Thus n ths model wage adjustment does not cure the labour market stuaton for unsklled workers. 21 Overall, relatve demand for sklled workers ncreases manly due to the unsklled labour-based techncal progress. 21 It should be noted, however, that n ths model, frst, we do not assume factor substtuton, thus ncreasng demand for unsklled workers, as they become relatvely cheaper, and second, there are no workers leavng the labour market. 23