Globalisation, Wages and Unemployment: A New Economic Geography Perspective Jolanda Peeters and Harry Garretsen 1

Transcription

1 Globalisation, Wages and Unemployment: A New Economic Geograpy Perspective Jolanda Peeters and Harry Garretsen 1 1. Introduction Tere is considerable cross-country variance in unemployment and employment performance as well as in te evolution of industrial and occupational structure. Differences are in particular striking between te US and Europe, but tere are also substantia l differences among European countries (Burda and Dluosc, 1998, p. 5). Tis quotation is from an interesting study tat tries to analyse te relevance of globalisation for te observed canges in wages and unemployment of low-skilled labour in Western industrialised countries. Te empirical evidence suggests tat tis relevance varies significantly across tese countries (OECD, 1997). In some countries globalisation, eiter troug increased trade or increased factor mobility, is tougt to ave ad an adverse impact on particularly te relative wages of lowskilled workers, wereas in oter countries relative low-skilled wages are ardly affected or ave even increased. Moreover, in some cases te impact of globalisation does not seem to sow up in relative wage canges but (also) in an increase in te relative unemployment of low-skilled labour (Dewatripont et al, 1999). One possibility to explain tese differences is tat tere may be considerable cross-country variation in te degree of globalisation as well as in te workings of national labour markets. Wit respect to te latter te well-known differences in wage flexibility between notably te US and te UK and te continental European countries come to mind (Layard et al, 1991). So, one would expect tat te teoretical model wic as been most widely used to analyse te possible impact of globalisation on labour markets to take tese differences into account. Tis is, owever, clearly not te case. Te model we refer to is, of course, te Heckscer-Olin model and in many discussions on globalisation te simple 2x2x2 textbook version of tis model is used (see for instance te contributions to Collins, 1998). Tis model makes unambiguous predictions about te impact of globalisation. Te empirical validity of tese predictions is not our concern ere (but see Peeters, 2001). Wat matters for our present purposes is tat tese predictions are based on a one size fits all analysis of globalisation. Countries are omogenous wit respect to te degree and contents of globalisation tat tey experience and also wit respect to te workings of te labour market (typically wages are fully flexible). 2 Given te aforementioned cross-country variation, tis is a rater unsatisfactory state of affairs. 1 At te time of te writing of tis paper bot autors were affiliated wit te Dept. of Applied Economics, University of Nijmegen, te Neterlands. We would like to tank Steven Brakman, Carles van Marrewijk. Albert de Vaal and participants at te conference (notably Peter Neary, Russell Cooper and our discussant Teo van de Klundert) for teir comments. Of course, te usual disclaimer applies. Tis capter is a revised version of our CESifo Working Paper no. 256, February 2000, Center for Economic Studies, Munic. 2 For exceptions in terms of a Heckscer-Olin framework see Krugman (1995) and Davis (1998).

2 One solution is to use an alternative trade model tat enables a ricer menu of possibilities as far as te impact of globalisation on labour markets is concerned. In our view and following Dluosc (2000) te new economic geograpy approac may offer suc an alternative. Krugman and Venables (1995) develop a two country new economic geograpy model to analyse te effects of globalisation on te relative wage between tese two countries. In teir model tere is, owever, only one type of transportation costs, one type of labour and te labour market always clears. Tis seriously limits te relevance of te model for te issue of globalisation and its impact on labour markets. In tis paper we build a igly stylised new economic geograpy model in wic two types of labour (low and igskilled) are distinguised and in wic, due to wage rigidities, unemployment may arise. Moreover, we also introduce two types of transportation costs (for goods and services respectively) to allow for a more intricate analysis of te ways in wic globalisation can influence te economy. In sum, te purpose of tis paper is to analyse te impact of globalisation, wic we mainly define as te fall in te transportation costs of goods and/or services, on te labour market position of low-skilled workers. By making use of a new economic geograpy model tat allows for wage rigidities, we expect to find part of te teoretical explanation for te diverging labour market experiences across te advanced countries in te globalisation era. Te paper will proceed along te following lines. Section 2 briefly discusses te model and focuses on its two most relevant and novel features, te production structure and te workings of te labour market. Section 3 investigates te sort-run equilibrium of te model, in wic labour cannot move between regions. As a first approximation tis is a reasonable assumption because te international mobility of labour is still relatively unimportant as a determinant of globalisation compared to te role of international trade and capital mobility (IMF, 1997). In te end, owever, we must also deal wit te implications of te mobility of labour (if only because te mobility of labour (and firms) distinguises te new trade teory from te new economic geograpy). Section 4 is terefore concerned wit te long-run implications of globalisation by allowing for te mobility of ig-skilled labour. Section 5 concludes. 2. Te Model 2.1 Main Ingredients of te Model Our model is based on te new economic geograpy model developed by de Vaal and van den Berg (1999) but tere are some notable differences between teir and our approac. Te main similarity is to be found in te modelling of te production structure (see below). Te main differences are tat we include two instead of only one factor(s) of production (ig- and low-skilled labour) and allow not only for flexible wages but also for rigid wages. 3 Apart from te andling of te production structure and te labour market, our model resembles te standard two region new economic geograpy model 3 In a twin paper, Peeters and de Vaal (2000), te same mo del is used but labour market rigidities are not part of te analysis. 1

3 as laid out in f.i. Fujita, Krugman and Venables (1999, capters 4 and 5) or Brakman, Garretsen and van Marrewijk (2001, capters 3 and 4). Demand Te world is divided in two regions, Home and Foreign. Bot regions are capable of producing a omogenous agricultural good A and a variety of te eterogeneous manufactured good M. Te agricultural good, wic serves as te numéraire, is produced wit a constant returns to scale tecnology, wereas te manufactured goods are produced under economies of scale. All individuals 1 are assumed to ave a Cobb-Douglas utility function of te form U = C µ C µ, were C M is consumption of te manufactures aggregate and C A is consumption of te agricultural good. Te manufactures aggregate is given by C = c + c θ θ M f N N f 1/ θ, were N is te number of varieties tat originate in Home and N f is te number of varieties tat originate in Foreign. As te manufacturing varieties are imperfect substitutes for eac oter (0 < θ < 1), te Home consumer consumes bot Home (c ) and Foreign (c f ) varieties. 4 Tis merely restates te crucial role of Dixit-Stiglitz model in new economic geograpy. Te relative demand is a function of relative prices only: (1) c c f = 1 1 p θ pf were p denotes te price of a Home good consumed by a Home consumer, wereas p f is te price of a Foreign good consumed by a Home consumer. Utility maximisation gives us in addition te true price index of consumers in Home, wic, as we will see below, is necessary to derive real wages: M A (2) θ 1 θ θ θ θ 1 θ 1 = + f f P% N p N p Supply Te demand side of our model is standard witin te new economic geograpy approac but tis is not te case for te production structure. Te production of manufactures in Home cons ists of two activities: te production of unfinised goods, x, and te adding of producer services from Home and Foreign, S and S f. Wile te production of unfinised goods is performed by te manufacturing firms, te production of producer services is performed by services firms. 5 It is assumed tat te unfinised goods are produced by low-skilled labour only and tat te production of unfinised goods entails 4 We describe te Home region, simply noting tat analogous conditions old in Foreign. 5 Tus, a manufacturing firm buys te producer services from te services firms. In tis respect, producer services can be regarded as an intermediate input to te manufacturing firms. 2

4 scale economies: L F Bx m x m = +, were x L is te low-skilled labour used to produce te output x and F (B) are te fixed (marginal) cost in terms of low-skilled labour. Part of te unfinised output of te Home manufacturing firm is intended for te Home region (x ), wereas te oter part is exported to Foreign (x f ); tus: x = x + xf. Producer services are produced by ig-skilled labour only and its production also entails scale economies: H f bs S = +, were HS is te ig-skilled labour used to produce services output S. As producer services typically require close interaction between te consumer and te producer, we assume tat producer services ave to be added locally. An example elps to clarify te production structure. Suppose a ome firm produces a variety of a particular manufactured good, let s say a car. Te car itself is te unfinised good (x ) and it requires te addition of producers services produced in Home and Foreign before it can be sold as a final consumption good to te consumers in bot regions. In te case of te car, te producer services tat are required are for instance all kinds of dealer, sales and financing services witout wic te consumption good can not be sold to te consumer. 6 Finally, for te location of production to matter, it must te case tat bot te trade of unfinised goods and producer services involves transportation costs. We assume te familiar iceberg transportation costs for unfinised goods (τ) and producer services (ρ), so tat a part of te transported item melts away. Te production functions of a Home manufacturing firm can ten be given by: (3) m = x S + ( ρs f ) n nf α 1 α (4) α m = ( τx ) ( ρs ) + S n nf f f f ff 1 α were m and m f denote te amount of te final consumption good a Home manufacturing producer makes for consumers in Home and Foreign; n and n f are te number of producer services varieties from Home and Foreign; S ijk refers to te use of a region i service by a region j manufacturing producer to finis its good in region k; is te elasticity of substitution of producer services and α is te unfinised good intensity in final production. Equilibrium Prices and Production 6 In a way we could also denote our two production factors as labour type A and type B instead of as low-skilled and ig-skilled labour. Te skill level as suc is not important in wat follows. 3

5 Given tis particular production structure wic will play an important part in our analysis, te determination of equilibrium prices and production is more or less straigtforward so we will be brief ere. Manufacturing producers maximise teir profits. Tis yields te following price equations 7 : (5) (6) p p C = wv% θ α 1 C = τ wv% θ α α 1 α f f wit C = B α α -α (1-α) -(1-α) > 0 and were w denotes te nominal wage of low-skilled workers in Home (te equilibrium level). Home given by: (7) V% v 1 1 f = v + n n ρ f V % i denotes te price index of producer services in region i =, f and is for (1 ) If tere is free entry and exit of firms into manufacturing, profits must be zero. Substitution of earlier results ten leads to te result tat te total production of unfinised goods and its use of lowskilled labour is alike for all manufactures trougout te economy and is equal to (per firm): (8) x = αθ F 1 (1 ) M θ α L 1 x = F θ B 1 θ Equilibrium on te market for unfinised manufactured goods requires tat total sales of one firm equals total demand (per firm). Hence: (9) x αθ p m + p m = B w f f Let us now turn to te prices and production of producer services. Producer services firms also maximise teir profits; tis yields: b (10) v = r were r is te nominal wage of ig-skilled labour in Home. Here also, free entry and exit of firms into te services market drives profits to zero and leads to te result tat te output and te labour use of producer services firms is alike for all services firms and is equal to (per firm): (11) S f f = HS 1 b = 1 Equilibrium on te services market requires total sales per firm (v S ) to equal total demand, so te services market equilibrium is represented by 8 : 7 For a derivation; see Peeters (2001, Appendix C). 4

6 (12) vs 1 ( N pm + N f pfmf) V% = (1 αθ ) v ρ ( N f pffmff + Npfmf ) V% f 1 Te model can be closed by specifying te equilibrium conditions for te labour market and te market for final consumption goods. Our specification of te labour market deviates, owever, from te standard new economic geograpy models in two important ways. First, our model includes two types of labour and, second, our labour market sould someow allow for wage rigidities. Te next sub-section terefore takes a closer look at te workings of te labour market in our model. 2.2 Te Labour Market and Wage Rigidity Recall tat tere are two factors of production in eac economy: low-skilled labour (L ) and igskilled labour ( H ). In Home as well as Foreign low-skilled labour is employed in bot te agricultural A M and te manufacturing sector and is assumed to be immobile between sectors or regions: L= L + L. Te low-skilled workers employed in te agricultural sector (L A ) are equally divided between te two regions: A Li 1 = ll (i =, f). Te low-skilled workers employed in te manufacturing sector (L M ) are 2 not necessary divided equally between Home and Foreign: λ l is te sare of low-skilled manufacturing workers in Home. Due to wage rigidities, to wic we turn below, te labour market for low-skilled manufacturing workers in Home does not ave to clear and unemployment can arise. If tis is te case,, U refers to te number of unemployed in Home, wile L refers to te number of employed lowskilled manufacturing workers in Home. Note tat in Foreign te labour market for low-skilled manufacturing workers always clears. Te ig-skilled workers in bot regions are always fully employed and tey are assumed to be mobile between Home and Foreign in te long run: λ is te sare of ig-skilled workers in Home. Substitution between low-skilled and ig-skilled workers is not possible. Te set-up of te labour market is given by equations (13) and (14): (13) NL M x (1 l) λll if flexible = (1 l) λll U if rigid Me M N L = (1 l)(1 λ ) L f xf l 8 To derive total demand, note tat eac services firm delivers to Home and Foreign manufacturing firms S = ( S + S f ) + ( S + S f ff ). From te FOC of profit maximisation of manufacturing firms, it N N f follows tat te demand for te output of a region i services firm by a region j manufacturing producer for consumers in region k is equal to: 1 S ijk θ % = = 1 1 (1 α) p m v V ( for i k) jk jk i k % 1 jk jk i k θ(1 αρ ) p m v V ( for i k). 5

7 (14) nh S = λ H nh = (1 λ ) H f Sf Equations (13) and (14) give labour demand (left-and side) and labour supply (rigt-and side). As can be seen, wit te notable exception of te low-skilled labour market wit rigid wages, te various labour markets clear. In case te labour market clears it is also immediately clear tat te equilibrium number of varieties is determined. Note, owever, tat tis is not true for N if U > 0. Te specification of te equilibrium conditions for te market of manufactured goods closes te model. Note tat te unemployed ave noting to spend in tis model wic inter alia means tat total income is always smaller wit unemployment. Given te Cobb-Douglas representation of te demand side, so tat consumers spend a fixed sare of teir income on bot goods, namely, µ on manufactured goods and (1-µ) on agricultural goods, te equilibrium conditions for te market for manufactures in Home and Foreign are respectively: (15) N p c f f f M A ( wl L rh ) rig Me, A ( w L + L + rh ) µ N p c = µ M A (16) N f pffcff + N pfcf = µ ( wl f f + Lf + rh f f ) were if flexible if rigid rig w is te absolute wage of low-skilled workers in Home if te labour market does not clear. Given te importance attaced to wage rigidity in our analysis of te effects of globalisation, it is useful to elaborate upon te implications of wage rigidity. We distinguis two labour market regimes in Home. Te first is te flexible (FLEX) regime, in wic te labour market for low-skilled manufacturing workers always clears (i.e. U equals zero). In fact, in tis regime our labour market is equivalent to te specification of te labour market in te core new economic geograpy model (see Fujita et al, 1999, p. 52). Te number of varieties N i and n i (wit i =, f) are ten determined on te labour market (see equations (13) and (14)), wereas te wages of low-skilled and ig-skilled workers are determined on te goods market (equations (15) and (16)). As long as te distribution of ig- and low-skilled workers between Home and Foreign is fixed, labour supplies are also fixed and tis gives te equilibrium number of varieties of unfinised goods and producer services. Our second labour market regime is te relative wage rigidity (RWR) regime, in wic it is assumed tat te low-skilled wages are set so as to equal te ig-skilled wages, ence w rig = w RWR = r. In fact, tis is a very stringent wage norm because it implies tat tere is no wage differential watsoever. For our analysis it does, owever, not make muc difference weter low-skilled wages are 50%, 75% or, as in our case, 100% of ig-skilled wages. Tis rigidity reflects a wage norm wic is tougt to be very relevant for f.i. te countries in continental Europe. Tis wage rigidity always binds (because w FLEX < r FLEX ) meaning tat part of te low-skilled manufacturing labour force in Home will be unemployed. To see ow unemployment arises, in case of relative wage rigidity (RWR) and in contrast wit te FLEX-regime, note tat N can no longer be looked upon as being determined on te 6

8 labour market. Instead, N is ten determined on te goods market. Wit w fixed at te level of te ig-skilled wage in Home and, also important, assuming tat N f, n and n f are still determined on teir respective labour markets, N as to adjust in order for te goods market (equations (15) and (16)) to continue to clear. Tis means tat te labour market for low-skilled in Home will not clear. Tis sows up in te rigid wage version of equation (13) were U represents te disequilibrium on tis labour market. In fact, it turns out tat wit relative wage rigidity, N as to decrease tereby reducing te number of varieties produced Sort-Run Implications of Globalisation To see ow globalisation affects te relative labour market position of te low-skilled in Home, we focus in tis section on te sort-run equilibrium and take terefore te distribution of low-skilled and ig-skilled labour between regions as given. We tink tat te sort-run equilibrium is interesting from a globalisation perspective because, as was stated in te introduction, te international mobility of labour is still relatively low. Tis means tat globalisation in terms of our model occurs troug te effects of a fall in transportation costs of goods and services between our two regions. Te sort-run equilibrium is contained in te equations (1), (5), (6), (9), (12), teir Foreign equivalents and equations (13) - (16). As wit most new economic geograpy models te complexity of our model precludes an analytical solution, so we will conduct our analysis troug simulations. Te numerical values for te bencmark case are given in Table 1. Table 1: Numerical values in te bencmark L = 800 F = 1 α = 0.6 H = 100 f = 0.5 µ = 0.6 l = 0.5 B = 0.05 θ = = 0.6 λ l = λ = 0.4 b = 0.1 τ = ρ = Relative Wages between Regions Tis section investigates te impact of globalisation (a reduction in transportation costs of unfinised goods and/or producer services) on te relative wages of low-skilled and ig-skilled workers between regions (w /w f and r /r f ). Since most new economic geograpy models ave only one factor of production tey can only address te impact of globalisation on te relative wage between regions (e.g. Krugman and Venables (1995)). Focusing first on relative wages between regions allows us, albeit in te context of a sort-run equilibrium, to compare our results wit te results of tese onefactor models. In sort, te standard conclusions (essentially based on Krugman, 1991, p. 491) can be stated as follows: -1- if bot regions are equally well endowed wit labour, te relative wage will 9 Of course, oter types of wage rigidity are also possible. In Peeters and Garretsen (2000) we also analyse te implications of a minimum wage for te low-skilled in Home. Te main conclusions are, owever, in tat case not different from te ones derived in sections 3 and 4 for te FLEX and RWR regimes. 7

9 equal one; -2- if labour is ten sifted from one region to anoter, te relative wage can move eiter way depending on te trade-off between te Home Market and Extent of Competition effect; -3- te level of transportation costs is very important in settling te balance between bot tese effects Figure 1: Te impact of a reduction in te transportation costs of goods on te relative wages of low-skilled workers between regions relative wages of low-skilled workers (Home/Foreign) transportation cost of goods lambda l = lambda = 0.4 lambda l = lambda = 0.5 lambda l = lambda = 0.6 In order to investigate weter tese effects are present in our model, consider Figure 1 wic sows te effect of a reduction in te transportation costs of unfinised goods (increase in τ) on te relative wage of low-skilled workers between regions (w /w f ) in te regime wit flexible wages. Tree distributions of labour are sown: te first one is simply te bencmark case, in wic Home is smaller in terms of low-skilled and ig-skilled labour (λ =λ l =0.4), te second one is te symmetrical case (λ =λ l =0.5), and te tird one is for Home being te larger region (λ =λ l =0.6). In te symmetrical case, w /w f indeed equals one at all levels of τ; as tere are no differences between te two regions all endogenous variables will ave te same values in Home compared to Foreign. In te case Home is te smaller region (λ =λ l =0.4), owever, te relative wage does not equal one and a fall in te transportation costs of unfinised goods implies tat w /w f decreases for low levels of τ. After a critical level of transportation costs (τ > 0.46), w /w f starts to increase again. Figure 1 sows tat if Home is instead te larger region (λ =λ l =0.6) exactly te opposite appens wit te relative wage w /w f. Te relation between transportation costs and te relative wage is tus non-linear; a fall in te 10 In a nutsell, te Home Market effect gives firms in te larger region a competitive advantage by allowing tem to better exploit economies of scale, wic enables tem to offer iger wages, wereas te Extent of Competition effect gives firms in te smaller region an advantage as tey face less competition for te local demand, wic enables tem to offer iger wages (see: Krugman, 1991). 8

10 transportation costs of unfinised goods can eiter increase or decrease te relative wage depending upon te level of transportation costs. Figure 1 tus allows us to reproduce te two oter aforementioned standard results. To see tis, note tat a sift of labour from one region to anoter, wat in Figure 1 boils down to comparing, for a given level of τ, te relative wage in te symmetrical case wit te two oter λ-distributions implies tat w /w f can move eiter way depending upon te level of τ. At low levels of τ, a sift of labour from Home to Foreign implies tat te relative wage of low-skilled workers in Home increases, as te Extent of Competition effect dominates over te Home Market effect. At iger values of τ, te balance between bot effects is reversed toug, and a sift of labour from Home to Foreign implies a fall in te relative wage of low-skilled workers in Home. Wit Home being te larger region te opposite appens. Of course, tis kind of analysis can be repeated for te relative wage of ig-skilled workers between regions (r /r f ) or for a reduction in te transportation costs of producer services (increase in ρ). Tis leads to similar conclusions. If one looks at absolute wages it is noteworty tat wages in te same region always move in te same direction. So if for instance w increases, so will r. Tis is to expected because te tug of war between te Home Market effect and te Extent of Competition effect as a similar qualitative impact on intra-regional wages. 11 By and large similar conclusions wit respect to relative wages between regions old in te RWR regime (not sown ere). Tere is one notable difference wit te case of flexible wages, namely, te relative wage need to be equal to one in te symmetrical case. Tus, altoug bot kinds of labour are equally divided between Home and Foreign, te absolute wages of te same type of labour differ between Home and Foreign. Tis can be explained by te fact tat unemployment exists in te case of relative wage rigidity and unemployment reduces te number of employed workers and terefore decreases te size of te Home region (because te unemployed ave, by assumption, noting to spend). Tis implies tat, altoug we assumed an equal distribution of labour, an asymmetry between te two regions exists so tat te relative wage does not ave to equal one. Finally, and again in accordance wit te core model of new economic geograpy, we know wat appens to te absolute wages in Foreign. Tey move in precisely te opposite direction from absolute wages in Home. Wen w and r increase, it is always te case tat w f and r f decrease by te same amount. Te upsot of Figure 1 is tat canges in transportation costs lead to canges in relative wages between regions and to canges in absolute wages as well tat can be understood in terms of te by now standard battle in new economic geograpy models between te Home Market and Extent of Competition effect. How tis battle is decided depends crucially on te λ-distribution between Home and Foreign and on te level of transportation costs. In wat follows is important to keep in mind tat tese two standard effects remain relevant wen transportation costs fall. 9

11 3.2 Relative Intra-Regional Wages and Unemployment Altoug section 3.1 gives rise to interesting (but well-known) results, te main interest of tis paper is te impact of globalisation on te relative wage of te low-skilled (w /r ) or te unemployment rate (U ) witin a region, ere te Home region. Tis sub-section investigates te distributional effects of globalisation witin te two regions in te sort run and ere we can compare our results wit te results found by more traditional models like te Heckscer-Olin model. In te streamlined 2x2x2 version of tis last model, wic as dominated te globalisation debate (see Collins (1998) or Wood (1998)), te effect of a reduction of transportation costs on te relative wages in te country wic is relatively well-endowed wit ig-skilled labour is clear-cut. Suc a reduction leads to a decrease of relative low-skilled wages. In te Heckscer-Olin approac, te supply side (= te distribution of endowments) drives tis result. In our model, te action is very muc, like in any new economic geograpy model, on te demand side of te economy. We first investigate te decline in te transportation costs of unfinised goods (i.e. increase in τ) in section and ten turn to te decline in te transportation costs of producer services (i.e. increase in ρ) in section Before doing so, it is important to make clear tat in addition to te familiar Home Market and Extent of Competition effect, tere are wit respect to intra-regional labour market developments two effects to take into account. Bot of tese two additional effects relate to te specifics of our model: 1. Altoug w i and r i always move in te same direction after a fall in te transportation costs (see section 3.1), our production structure implies tat te absolute wages react quantitatively different to a cange in eiter τ or ρ. In particular, a cange in te transportation costs of unfinised goods (τ) always as a relatively larger impact on low-skilled wages, wile a cange in te transportation costs of producer services (ρ ) is more important for ig-skilled wages. 12 Tis result is due to te fact tat an increase in τ (ρ) is of direct relevance to te low-skilled (ig-skilled) as tey are te only factor of production used in te unfinised goods (producer services) production. Tis is of course relevant for te determination of te relative wage since any cange in w /r is simply a result of te canges in w and canges in r. Given tat an increase in τ implies a larger cange in w tan in r (and also tat te opposite olds wit respect to an increase in ρ), it is possible to understand te developments in w /r, even toug te absolute wages are affected (qualitatively) in te same way. Of course, tis is a partial line of reasoning and tere are oter intricate ways in 11 Tis raises te question as to ow globalisation can lead to canges in relative regional wages, tis question will be addressed in section To see tis, note from equations (6) and (9) tat an increase in τ as a direct impact on w. Also, from equations (10) and (12) it can be discerned tat an increase in ρ as a direct impact on r. Te finding tat w i (r i ) is affected quantitatively more by a cange in τ (ρ) is supported by te simulations. 10

12 wic canges in transportation costs influence wages in our model but tis line of reasoning is supported by te simulations. 2. If wages are no longer flexible (in our case: a fixed relative wage in Home), tis obviously also creates a cannel toroug wic intra-regional labour markets are affected. Tis is in particular true in Home were, as we will see below, te wage rigidity gives rise to unemployment. We now turn to te simulation results wic are tus te summation of te standard new economic geograpy effects of section 3.1 and te two above mentioned effects. Given te complexity of te model we tus ave to rely on simulation results and it will (unfortunately) often be impossible to fully disentangle tese effects for specific simulation results. Tis is a problem wit te new economic geograpy approac in general (Neary, 2001) Fall in te Transportation Costs of Unfinised Goods (increase in t) Figure 2 depicts for Home te relative wage of low-skilled workers and te unemployment rate respectively for te bencmark case were λ l = λ = 0.4 (Home is tus te smaller region). Te leftaxis displays te relative wage w /r and is relevant for te FLEX-regime. Te rigt-axis displays te unemployment U and tis variable is of interest in te RWR regime. Te figure immediately reveals one important conclusion: te relationsip between te transportation costs of unfinised goods and te relative wages of low-skilled workers and unemployment is non-linear and it does depend very muc on te level of τ wat te impact of globalisation on labour market in Home will be. Tis result sets our analysis apart from te traditional Heckscer-Olin conclusion tat a fall in transportation costs leads to a clear-cut cange in relative wages. Tat is to say, in te latter te cange in w /r does not depend on te degree of globalisation (= te level of transportation costs). Hence, te stylised fact tat industrialised countries differ considerably in terms of te canges in relative wages or unemployment in te globalisation era may tus ave someting to do wit differences in te level of transportation costs across countries. 11

13 Figure 2: Te impact of a fall in te transportation cost of goods on te labour market position of low-skilled workers relative wage of low-skilled workers (FLEX regime) transportation cost of goods unemployment (RWR regime) w/r if flexible (left axis) Consider for instance w /r and ence te FLEX regime in Figure 2: at low levels of τ (i.e. ig transportation costs), an increase in τ is bad news for low-skilled workers as tey face a decline in te relative wage; owever, after a certain level of τ (τ > 0.47 to be precise), a furter decline in te transportation costs of unfinised goods ceases to ave muc impact (in % cange). In fact, for low levels of transportation costs (ig τ) te relative wage increases sligtly. Te same applies in te RWR regime, were te labour market deterioration of te low-skilled at low levels of τ is now reflected in rising unemployment. Te main point to notice for bot regimes is tat te simulation results in Figure 2 sow tat te impact of globalisation on te labour market position of low-skilled workers depends crucially on te level of transportation costs. 13 A similar conclusion was reaced in case of relative wages between regions. As we argued in section 2, te disequilibrium in te labour market for low-skilled workers in Home is caused by te fall in te number of unfinised goods varieties in Home (N ) as te latter are now determined on te goods market. Tis implies tat te number of unfinised goods varieties is larger in te FLEX regime as compared to te RWR regime. Our simulation results sow tat tis is FLEX RWR indeed te case: N > N. As an aside to te existence of unemployment it can be added tat globalisation invariably leads to deindustrialisation in Home if te fall in transportation costs implies 13 In order to test te robustness of tis result, we performed an extensive sensitivity analysis. In most cases, te non-linear relationsip between τ and w /r is found. In addition we found tat te distribution of labour between regions is also crucial to determine te impact of globalisation on te labour market position of low-skilled workers in Home. So if we assumed tat Home was te larger region in terms of low-skilled and ig-skilled labour, globalisation turned out to be good news for te low-skilled at low levels of τ and bad news at iger levels of τ. We return to te distribution of labour between regions in section 4 below. 12

14 unemployment. Unemployment of low-skilled manufacturing labour in Home decreases te sare of te manufacturing goods sector (vis a vis te agricultural sector) in employment and production. Finally, note tat te fall in te transportation costs of unfinised goods as also implications for FLEX RWR FLEX RWR Foreign workers (not sown ere). It turns out tat invariably: w f < w f and r f < r f. 14 How can tis be explained? Note tat te wage rigidity in Home increases te absolute wages of low-skilled workers in Home. Te zero profit condition ten implies tat te prices of Home varieties will increase (see equation (5) and (6)), wic, by equation (1), decreases total demand for Home varieties and increases total demand for Foreign varieties. Te latter ten translates into iger absolute wages for low-skilled workers in Foreign. Tis conclusion for wages in Foreign supports te affirmative answer given by Davis (1998) to te question Does European Unemployment Prop Up American Wages? and it underscores is plea for a general equilibrium analysis of te impact of wage rigidities. Interestingly enoug, Davis (1998) reaced tis conclusion wit a Heckscer-Olin model but it tus carries over to our new economic geograpy model. Unemployment in Home is good news for all workers in Foreign Fall in te Transportation Costs of Producer Services (increase in r) Figure 3 depicts te relative wage of low-skilled workers and te unemployment rate, respectively, for Home in te two labour market regimes wen ρ increases (again te left-axis (rigt-axis) displays te relative real wage (low-skilled unemployment) wic is our focus in te FLEX (RWR) regime). From Figure 3 it can be seen tat tere is also a non-linear relationsip between te transportation costs of producer services and te relative wage of low-skilled workers. Te main message is tat if globalisation occurs troug a fall in te transportation costs of producer services, te relative wages of te low-skilled labour sow a downward trend for very low transportation costs (to be more precise, for ρ>0.63), but for iger transportation cost, a fall in te transportation cost of services is good news for te low-skilled in Home. Tis result can be compared wit te corresponding simulation result as sown by Figure 2. It ten turns out tat for te FLEX regime it may matter a great deal for te relative wage in Home weter τ or ρ canges. Tis leads us to conclude tat for te impact of globalisation one must also look at te type of transportation costs tat is being reduced. Our way to account for tis conclusion is given by point 1 at te end of te introduction to section 3.2. Tere we argued tat a cange in τ as a larger impact on low-skilled labour and te same for ρ and ig-skilled labour. Given tat an increase in τ implies a larger cange in w tan in r (and also tat te opposite olds wit respect to an increase in ρ), it is possible to understand te deviating developments in w /r for te case of flexible wages wen comparing te relative wages in Figures 2 and Recall tat te labour market always clears in Foreign. Te superscript in te notation of foreign wages relates to te labour market regime in Home. 13

15 relative wages of low-skilled workers (FLEX regime) Figure 3: Te impact of a fall in te transportation cost of services on te labour market position of low-skilled workers transportation cost of services w/r if flex (left axis) Te simulation result in Figure 3 for te relationsip between unemployment and te fall in te transportation costs of producer services tell a similar story. For te RWR regime it also turns out tat for unemployment te consequences of globalisation depend on te type of transportation costs wic is being reduced. Te differences in te outcome for unemployment in case τ or ρ canges, can be traced back to te efficiency or productivity effect related to a fall in te transportation costs of producer services (see equations (3) and (4)). An increase in ρ implies tat less services are being lost in international trade; implicitly, tis increases te number of producer services available per lowskilled worker and tus te productivity of low-skilled workers rises. An increase in τ does not ave tis effect, it merely reduces te transportation costs of unfinised goods between Home and Foreign witout affecting te productivity of te factors of production unemployment (RWR regime) 4. Long-Run Implications of Globalisation We finally turn to te impact of globalisation in te long run and introduce a second aspect of globalisation: factor mobility. In te new economic geograpy models, te interregional mobility of labour differentiates te long run from te sort run. 15 So far, neiter low-skilled nor ig-skilled workers were mobile. Tis assumption is partly relaxed in tis section because it is now assumed tat ig-skilled workers can move between Home and Foreign and tat low-skilled workers remain immobile. Tis is a reasonable assumption given tat ig-skilled workers are typically more mobile tan low-skilled workers (e.g. Siebert, 1999) and te still very limited mobility of low-skilled labour in Europe (OECD, 1997). As is common in te new economic geograpy literature, it is assumed tat 14

16 ig-skilled workers move until eiter real ig-skilled wages are equalised or until all ig-skilled workers moved to one region. In order to investigate te long-run equilibrium, we first derive te relative real wage of igskilled workers (Home/Foreign). Real wages of ig-skilled workers in region i =, f are a function of nominal wages and prices as follows: υ i = ri / P % µ i, were P % i is te price index as given in Equation (2). Now substitute te price equations in te price indices and define te relative real wage of ig-skilled workers as υ /υ f wic is ten given by: (17) υ υ r αθ αθ w θ 1 θ 1 N f wf + N (1 α ) µ τ V% θ 1 f = 1 f r f V% θ αθ θ αθ w θ 1 θ 1 f Nw N f + τ Given te relative real wage of ig-skilled workers, we are interested in te following questions. Wat are te caracteristics of te long-run equilibria and, related to tis, ow are tese equilibria affected by globalisation? 16 θ µ 1.2 Figure 4.a: Te long run equilibrium for lambda l = 0.4 relative real wages of ig-skilled workers (Home/Foreign) lambda tau = ro = 0.2 tau = ro = It also differentiates te new economic geograpy from te new trade teory (see Neary, 2001). In tis sense te discussion so far (wit λ fixed) as even more in common wit for instance Krugman and Venables (1990) tan wit Krugman (1991) and te literature tat builds on te latter. 16 Anoter question is, of course, ow in te long run relative (real) wages witin a region cange as λ, ρ or τ cange. Since te simulations (not sown ere) sow tat te conclusions as to te beaviour of relative (real) wages in te long run are comparable as for te relative intra -regional wages in te sort-run, in te sense tat te labour market outcomes depend crucially on te level and type of transportation costs being reduced, te distribution of labour across regions and te labour market regime, we do not address tis question in our paper. 15

17 Figure 4.b: Te long run equilibrium for lambda l = relative real wage of ig-skilled workers (Home/Foreign) lambda tau = ro = 0.2 tau = ro = 0.5 Figure 4a sows te computed values of υ /υ f as a function of λ for two different levels of transportation costs and wit λ l =0.4 in te regime wit flexible wages. Figure 4b does te same but now for λ l =0.6. It is clear tat te long-run outcome for υ /υ f and ence for te equilibrium distribution of ig-skilled labour depends on te level of transportation costs and te fixed distribution of low-skilled labour. For ig transportation costs (τ=ρ=0.2), tere is a stable intermediate equilibrium for λ. However, tis intermediate equilibrium ceases to exist if transportation costs fall. Wit τ=ρ=0.5, te mobile labour force ends in eiter Home (Figure 4b) or Foreign (Figure 4a) depending on te value of λ l. Te relevance of λ l is clear wen comparing Figures 4a and 4b, if a region as more low-skilled labour it ceteris paribus ends up wit more ig-skilled labour. Simulations like tose sown in Figure 4 for te relative wage rigidity regimeyield by and large similar conclusions 17. Te sift from a stable intermediate equilibrium to a stable monocentric equilibrium (λ =0 for λ l =0.4 and λ =1 for λ l =0.6) is in line wit te standard new economic geograpy result based on Krugman (1991). Te reason beind tis result is tat below some level of transportation costs, te Home Market and, now also, te Price Index effect dominate te Extent of Competition effect. 18 Hig- 17 Tere is one big difference between te FLEX and te RWR regime. In te RWR regime, te stable intermediate equilibrium is reaced only if Home is assumed to be te larger region (λ l =0.6) and transportation cost are sufficiently ig (τ=ρ=0.2); if Home is te smaller region (λ l =0.4), te stable intermediate equilibrium ceases to exist even for τ=ρ=0.2. For an explanation, see Peeters (2001, pp ). 18 Te Price Index effect implies tat te price index wic consumers face will be lower in te larger region. Consumers/workers in te larger region ave te advantage tat tey pay a relatively lower price for te final 16

18 skilled workers will terefore all move to te larger region and complete agglomeration of ig-skilled workers will be te outcome. For ig transportation costs, tis order is reversed, and, in our model, te relative strengt of te Extent of Competition effect ensures a stable spreading equilibrium for ig-skilled workers. 5. Conclusions In tis capter, we ave developed an alternative model to analyse te effects of globalisation. In tis model, essentially a two factor new economic geograpy model wit te possibility of relative wage rigidity, te impact of globalisation (proxied by a fall in transportation costs and labour mobility) on te labour market varies considerably. A first conclusion is tat te impact of globalisation on relative wages and unemployment turns out to depend critically on te level and type of te transportation costs tat is being reduced, te flexibility of wages and te distribution of low- and ig-skilled labour across countries. Globalisation is by no means always bad news for te low-skilled. On te contrary, in some of te simulations we performed, te low-skilled are better off in terms of te relative wage and/or te unemployment outcome as globalisation proceeds. Because te OECD countries differ in all tese factors, our model can give a teoretical account for te diverging labour market outcomes across te OECD countries. A second conclusion relates to te difference between te sort run and te long run in tis type of model. In te sort run, tus for a given distribution of labour, globalisation typically as a larger (positive or negative) impact in its initial stages. In te long run, toug, in wic globalisation proceeds bot by te mobility of labour, it is difficult to say wen globalisation implies a cange in te agglomeration pattern. A final conclusion is tat our analysis illustrates tat it can be very misleading to analyse te effects of globalisation based on a model wit fully flexible wages for countries in wic wage rigidity prevails. It matters a great deal for te effects of globalisation weter or not wages are rigid. Our analysis points to two promising avenues for furter researc. Te first one is to try to substantiate te teoretical conclusions wit empirical evidence. Te second one is to extend te teoretical model by analysing te consequences of productivity differences between sectors and regions in order to deal wit tis oter alleged villain in te globalisation debate, (skill biased) tecnological cange. References Baldwin, R.E. and R. Forslid (1997). Te Core Peripery Model and Endogenous Growt, CEPR Working Paper, no Brakman, S., H. Garretsen, C. Van Marrewijk (2001), An Introduction to Geograpical Economics: Trade, Location and Growt, Cambridge University Press, Cambridge. Burda, M. C., and Dluosc, B. (1998). Globalization and European Labour Markets. CEPR Discussion Paper, no Collins, S.M. (1998). Imports, Exports and te American Worker, Brookings Institution Press, Wasington D.C. manufacturing good tey consume and tis advantage sows up as a relatively iger real wage tus enancing te attractiveness for mobile workers to move/stay in te larger region. 17

19 Davis, D. R. (1998). Does European Unemployment Prop Up American Wages? National Labor Markets and Global Trade. American Economic Review, 88(3), Dewatripont, M., Sapir, A., and Sekkat, K. (1999). Trade and Jobs in Europe: Muc Ado about Noting?, Oxford University Press, Oxford. Dluosc, B. (2000). Industrial Location and Economic Integration, Centrifugal and Centripetal Forces in te New Europe, Edward Elgar, Celtenam. Fujita, M. Krugman, P. and A.J. Venables (1999). Te Spatial Economy: Cities, Regions and International Trade, Te MIT Press, Massacusetts. IMF, (1997), World Economic Outlook, May, Wasington. Krugman, P. R. (1991). Increasing Returns and Economic Geograpy. Journal of Political Economy, 9(3), Krugman, P. R. (1995). Growing World Trade: Causes and Consequences. Brookings Papers on Economic Activity, 1: Krugman, P.R. and A.J. Venables (1990), Integration and te Competitiveness of Periperal Industry in C. Bliss and J.B. de Macedo (eds.), Unity wit Diversity in te European Community, Cambridge University Press, Cambridge, pp Krugman, P. R., and Venables, A. J. (1995). Globalization and te Inequality of Nations. Quarterly Journal of Economics, 110(4), Layard, R., Nickell, S., Jackman, R. (1991). Unemployment: Macroeconomic Performance and te Labour Market, Oxford University Press, Oxford. Neary, J.P. (2001). Of Hype and Hyperbolas: Introducing te New Economic Geograpy, Journal of Economic Literature, vol XXXIX, pp OECD. (1997). Trade, Earnings and Employment: Assessing te Impact of Trade wit Emerging Economies on OECD labour markets. Employment Outlook (capter 4), OECD, Paris. Peeters, J. and A. de Vaal (2000). Explaining Te Wage Gap: Heckscer-Olin, Economic Geograpy and Services Availability SOM researc report, 00C21, University of Groningen. Peeters, J and H. Garretsen. (2000), Globalisation, Wages and Unemployment: A New Economic Geograpy Perspective, CESifo Working Paper, no. 256, Munic. Peeters, J. (2001), Globalisation, Location and Labour Markets, PD-tesis, University of Nijmegen. Siebert, H. (1999). Te World Economy, Routledge, London and New York. Vaal, A. de and M. van den Berg (1999), Producer Services, Economic Geograpy, and Services Tradability. Journal of Regional Science, Vol. 39, No. 3, pp Wood, A. (1998), Globalisation and te Rise in Labour Market Inequalities, Economic Journal,