Trade in Services and Market Structure

Transcription

1 Trae n ervces an Market tructure Inaugural-ssertaton zur Erlangung es akaemschen Graes enes oktors er Wrtschafts- un ozalwssenschaften er Chrstan-Albrechts-Unverstät zu Kel Vorgelegt von Aan Islyam [Master of Arts n Economcs (KIMEP)] aus Almaty, Kasachstan. Kel, 202

2 Geruckt mt Genehmgung er Wrtschafts- un ozalwssenschaftlchen Fakultät er Chrstan-Albrechts-Unverstät zu Kel ekan: Prof.r. Roman Lesenfel Erstberchterstattener: Prof.r. Johannes Bröcker Zwetberchterstattener: Prof. r. Rolf Langhammer Tag er Abgabe er Arbet: 3 Februar, 202 Tag er münlchen Prüfung: 30 Ma, 202

3 Contents Contents... Acknowlegements... v Lst of abbrevatons... v Introucton hort Bblography... 3 Chapter One: Trae n Intermeate Proucer ervces uner Imperfect Competton Introucton Moel settng Partal ervatves of markup ata n benchmark an moel calbraton Results of the numercal experments Results of the base scenaro tablty of the ntal equlbrum umerare problem Other specfcatons of the moel Conclung remarks Bblography Appences Chapter Two: On Equlbrum tablty n the CGE moels wth Monopolstc Competton Introucton Moel structure Base moel settng Equatons of the moel n the mxe complementarty problem format ata an calbraton Results from the base moel an equlbrum stablty Interpretaton of the results Equlbrum stablty an unqueness Alternatve perspectve on stablty Analytcal expresson of stablty contons Alternatve moel formulaton Contons of full specalzaton n the IR sector Concluson Bblography Appences... 6 Chapter Three: FI n strbuton ervces an Trae versus Investment Trae-Off Introucton... 67

4 3.2. Moel settng Case : o nvestments Case 2: FI n manufacturng Case 3: FI n strbuton servces Case 4: FI n both manufacturng an strbuton servces Analytcal results etermnants of FI n strbuton etermnants of FI n manufacturng umercal results electng parameter values A non-monotonc relatonshp between trae costs an FI n strbuton Conclung remarks Bblography Appences...95 Concluson 06 Affrmaton 09 v

5 Acknowlegements Ths research has been supporte by the Germany Acaemc Exchange ervce (AA), Open ocety Insttute (OI) an the Unversty of Kel. I am grateful to Prof. Bröcker an Prof. Langhammer for research supervson. I apprecate all the comments an suggestons I receve at the conferences n Venna, Ax-en-Provence an Rome. I am also grateful to my famly, frens an colleagues n Kel an Almaty. v

6 Lst of abbrevatons CR CE CGE ERP FI GAM GAT GP IR MOAP MCM MCP EG AM UCTA VA WTO Constant returns to scale Constant elastcty of substtuton Computable general equlbrum moel xt tgltz Effectve rate of protecton Foregn rect Investments Generalze Algebrac Moelng ystem General agreement on trae n servces Gross omestc prouct Increasng returns to scale Marshallan output ajustment process Mcro consstency matrx Mxe complementarty problem ew Economc Geography ocal accountng matrx Unte atons Conference on Trae an evelopment Value Ae Worl Trae Organzaton v

7 Lst of Tables Table.. Results of servces trae lberalzaton n percentages... 2 Table.2. Results of servces trae lberalzaton n percentages for hgher markup... 5 Table.3. Results of servces trae lberalzaton n percentages... 6 Table.4. Results of servces trae lberalzaton n percentages for fferent numerare values... 8 Table.5. Results of servces trae lberalzaton n percentages Table.C. Results of servces trae lberalzaton n levels Table.. Mcro-consstency matrx (MCM) Table 2.. Mcro-consstency matrx Table 2.2. Percentage change from the benchmark after a tax cut (t = 0) Table 2.3. Percentage change from the benchmark after a tax cut for the close verson of the moel (t = 0) Table 2.4. Percentage change from the benchmark after a tax cut an fferent benchmark value of the IR sector output Table 2.A. Percentage change from the benchmark after tax cut (t = 0)... 6 Table 2.2A. Mcro-consstency matrx of the mofe M62 moel Table 3.. The rato of exports from the home to host country of the multnatonal wth an wthout FI n strbuton for fferent values of K, a an b Table 3.2. The value of the strbuton threshol for fferent parameter values v

8 Lst of Fgures Fgure.. The mpact of the elastcty of substtuton between varetes on markup epenng on the values of conjectural elastcty an the number of frms... 0 Fgure.2. The share of the outputs of fnal goos for fferent values of the prce share (An unstable ntal equlbrum case)... 5 Fgure.3. The share of the outputs of fnal goos for fferent values of the prce share (A stable ntal equlbrum case)... 7 Fgure.A.The proucton structure of the Y nustry Fgure 2.. Moel stablty Fgure 2.2. Moel stablty from EG perspectve: Intal scenaro Fgure 2.3. Moel stablty from EG perspectve: Fgure 2.4. Moel stablty from EG perspectve: Fgure 2.5. Moel stablty from EG perspectve: Fgure 2.6. Moel stablty from EG perspectve: The case wth mport tarffs on both goos Fgure 2.7. ynamc tme path of the relatve output of goo X Fgure 2.8. ynamc tme path of the relatve wage of sklle labor n X sector Fgure 2.B.ynamc path of the output of goo X Fgure 2.2B.ynamc path of the output of goo Z Fgure 2.3B.ynamc path of the number of frms Fgure 2.4B.ynamc path of welfare Fgure 2.5B.ynamc path of Fgure 2.6B.ynamc path of the unsklle labor wage Fgure 2.7B.ynamc path of the consumer expenture Fgure 2.8B.ynamc path of the frm level prce n nustry X Fgure 3.. Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an fxe costs of FI n strbuton (F) Fgure 3.2. (F = 0.6) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an blateral tarff rate (t) Fgure 3.3. (F = 0.4) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an southern tarff rate (t ) Fgure 3.4.A (FM = 0.3) an Fgure 3.4.B (FM = 0.35) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n strbuton (F) an southern tarff rate (t ) Fgure 3.5 (F = 0.457) The nvestment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an southern tarff rate (t ) Fgure 3.6 The ervatve of the fference n profts between moes 3 an for fferent values of the southern tarff rate (t )... 9 Fgure 3.A. trateges of the multnatonal frm v

9 Introucton The servce sector represents a substantal an rsng share of output, employment an FI n many countres. Ths makes t crucal for future growth an evelopment prospects. It s observe that the share of servces n value ae an employment s ncreasng wth the level of evelopment: startng from 35% of GP n the lowest ncome countres an up to over 70% n evelope countres (Hoekman, 2006). At the same tme, trae n servces consttutes only one ffth of the total worl trae (Hoekman, 2006). Ths coul be explane by the ntangble nature of many servces, by trae barrers, an n partcular by the fact that the servce trae wth ts four fferent moes of supply ( cross-borer trae, consumpton abroa, foregn afflates trae n servces through commercal presence, an temporary movement of persons) s hghly nsuffcently recore. Recore trae s bascally cross-borer trae. It s ths trae whch has grown faster than trae n goos as a result of the avancement of nformaton technology. ue to hgh relatvely to goos trae barrers, t s expecte that global welfare gans to be acheve after servces trae lberalzaton coul be substantal. In partcular, the servces trae s expecte to brng sgnfcant benefts to evelopng countres, where barrers are the hghest. The objectve of the ssertaton s to contrbute to the exstng knowlege n the fel of the servces trae an ts mpact on evelopng countres. It s also to nvestgate the mplcatons for trae polcy analyss of enhance ways of moelng of the nherent features attrbutable to the servces sector such as mperfectly compettve market structure. The most straghtforwar ncatons of the mperfectly compettve market structure n servces are the ntrnsc heterogenety of servces an the presence of ncreasng returns to scale. The ssertaton concentrates on two types of servces: proucer an strbuton servces. Both sectors play a key role n etermnng the qualty an prce of goos. In partcular, proucer servces as ntermeate servces are an mportant nput n the goos proucton that nfluences prouctvty of manufacturng goos. On the other han, the strbuton servces consttute an mportant lnk between proucers an fnal consumers. Consequently, ths ssertaton concentrates on stuyng the role of both types of servces an ther mpact on the economy wthn the vertcal proucton structure. The frst chapter of the ssertaton enttle Trae n Intermeate Proucer ervces uner Imperfect Competton buls upon the general equlbrum moels n Markusen et al. (2005) an Konan an Assche (2007) to analyze the mpact of servces trae lberalzaton uner olgopoly competton. It ams to stuy the mpact of trae n proucer servces on the welfare, ownstream nustry output, prces of the factors of proucton an the pattern of trae. Olgopoly market structure wth frms that make conjectures about the proucton of ther omestc an foregn rval frms s evelope. Moels wth constant elastcty of substtuton proucton functons capture specal features attrbutable to servces such as heterogenety an preference for varety. It s foun that large fferences n markups between the omestc an foregn frms coul lea to a negatve welfare effect. The secon chapter s enttle On Equlbrum tablty n the CGE moels wth Monopolstc Competton. A smplfe moel s constructe to stuy the effects that lea to the emergence of an unstable equlbrum an counterntutve comparatve statcs result. The

10 Global Corresponence Prncple an the ynamcs unerlyng the ntally statc moel are apple to stuy the moel stablty. We assume that there are two sectors of the economy: ncreasng returns to scale sector (IR) an constant returns to scale sector (CR). It s analytcally emonstrate that the elastcty of substtuton between varetes an the fference n the skll ntenstes of the two sectors of the economy wll ncrease the stablty of the ntal equlbrum. Moreover, t coul be prove that an a valorem tax cut wll have the same mpact on the recton of change of the key enogenous varables of the moel rrespectve of the unerlyng ata as gven n the socal accountng matrx (AM). The ynamc tme paths of the enogenous varables from unstable to stable equlbrum are also presente. Even though the moel presente n chapter 2 s just a reuce verson of the moel presente n chapter, t s mportant for explanng the results of the moel wth olgopoly competton. In other wors, t s mpossble to gve a complete nterpretaton of the results n chapter an make a jugment about the stablty of the ntal equlbrum wthout analyzng the smplfe verson of that moel presente n chapter 2. mlarly, the smplfe moel allows us to sngle out the effects that etermne the results of the moel an the role of the market structure. Moreover, no comprehensve analytcal results coul be presente for the moel n chapter as t coul be for the moel n chapter 2. Fnally, the thr chapter enttle FI n strbuton ervces an Trae versus Investment Trae-off stues the etermnants of FI n strbuton usng a partal equlbrum moel. Foregn rect Investments represent a form of moe 3 of servces trae. Ths s an mportant channel of servces trae especally for the strbuton sector whch manly comprses retal an wholesale trae. fferently from franchsng those two types of strbuton servces coul only be supple by FI. The moel represents an extenson of the stanar proxmty/concentraton traeoff an s motvate by evence n Hanson et al. (200) accorng to whch foregn rect nvestment coul be fferentate as proucton orente an strbuton orente. Consequently, a moel n whch a vertcally ntegrate multnatonal frm has several optons of foregn market penetraton such as exportng, FI n manufacturng an FI n strbuton s evelope. There s a market seekng motve to nvest n strbuton servces an there s a cost-orente motve to nvest n manufacturng. It s foun that market sze n the host country, prouctvtes of proucng fferentate goos, substtutablty an the host country strbuton margn ncrease ncentve for FI n strbuton. On the other han, transportaton cost an wage rates n both the host an the home country of the multnatonal weaken the ncentve of a multnatonal to nvest n strbuton servces. mlarly, market sze n the south an n the north, substtutablty, wage n the home country, prouctvty of proucng outsource goo n the south an the host country tarff wll ncrease ncentve to nvest n manufacturng. Moreover, t was foun numercally that there s a non-monotonc relatonshp between trae costs an FI n strbuton. The results of the ssertaton coul have some mplcatons for economc polcy makng. In partcular, t s always mportant for host country governments to ece not only whether foregn servces provers shoul be taxe or not but also whch nstruments shoul be use to acheve greater welfare gans. In the frst chapter of the ssertaton there are two polcy tools at the sposal of the host country government: per unt of output tarff or lump sum entry tax. It was foun that both are napproprate an free trae n proucer servces brngs hgher welfare. 2

11 Many evelopng countres use specal polces to attract FI. There are fferent polcy nstruments at the sposal of a host country government such as fscal ncentves n terms of nrect subses, tax releve an so on, fnancal ncentves n the form of government grants or crets at subsze rates an all other ways of preferental treatment (avarett an Venables, 2004). The reasons to attract FI coul be base ether on welfare mprovng conseratons or they coul be motvate by poltcal economy processes. Irrespectve of the reasons, knowng the etermnants of FI n the specfc servces sector such as strbuton coul facltate the mplementaton of host country polces. In the thr chapter of the ssertaton such etermnants are presente. ue to the vast emprcal an theoretcal lterature on the relatonshp between FI an trae, trae polces coul also have an mplct mpact on ncentves for FI. Ths s alreay acknowlege by polcymakers. In ths ssertaton we are tryng to stuy how ncentves to nvest n strbuton are affecte by the presence or absence of FI n manufacturng. hort Bblography Hanson, G.H., Matalon, R.J., laughter M.J., (200) Expanson trateges of U Multnatonal Frms Brookngs Trae Forum, pp Hoekman, Bernar. (2006). Lberalzaton of Trae n ervces: A urvey, Worl Bank workng paper. Konan,., Assche, A., (2007). Regulaton, Market tructure an ervce Trae Lberalzaton, Economc Moelng 24 (2007), pp Markusen, J., Rutherfor, T., an Tarr,., (2005) Trae an rect nvestment n proucer servces an the omestc market for expertse, Canaan Journal of Economcs 38 (2005), pp avarett, G., an Venables, A., (2004) Multnatonal Frms n the Worl Economy, Prnceton an Oxfor: Prnceton Unversty Press. 3

12 Chapter Trae n Intermeate Proucer ervces uner Imperfect Competton In ths chapter a stylze CGE moel s constructe to stuy the mpact of lberalzaton of barrers for foregn provers of ntermeate proucer servces uner mperfect competton on welfare, ownstream nustry output, prces of factors of proucton an the pattern of trae. An attempt s mae at ncorporatng olgopoly market structure nto the servces sector wthn a general equlbrum moel. Consequently, a moel wth frms makng output conjectures about omestc an foregn rvals s aopte. The case of a small evelopng country wth a less effcent servces sector relatve to foregn frms s assume. In ths framework, nteracton an the relatve sgnfcance of mechansms resultng from a preference for varety, pro compettve an effcency effects s analyze. It s foun that the lberalzaton of trae n servces mght be negatve n terms of welfare only f there s a sgnfcant fference n the relatve economes of scale an versfcaton between omestc an foregn frms. On the other han, unerlyng market structure s foun to have lttle mpact on results... Introucton There are many stues that emphasze the potental global gans from servces trae lberalzaton. Those gans are expecte to be hgher than the gans from trae n goos (Hoekman, 2006). In partcular, the servces trae s expecte to brng sgnfcant benefts to the evelopng countres where barrers are the hghest. The outcomes coul be mprovements n househol welfare (Rutherfor et al., 2006), long run growth performance (Mattoo et al., 2006) an omestc nustry prouctvty (Markusen et al., 2005). On the other han, numerous quanttatve moels show that the effects are uneven an some countres may even lose ue to varous reasons: rents accrung to foregn nvestors, terms of trae eteroraton, etc. It s also observe that the gans from FI flows whch coul be classfe as lberalzaton n terms of GAT (General Agreement on Trae n ervces) Moe 3 commercal presence an ownershp restrctons are large an more varable compare to the gans comng from the cross borer trae moe of supply (Hoeckman, 2006). mlarly, the mpact on the specfc factor such as sklle labor coul be more pronounce. The types of servces uner conseraton are typcally proucer servces such as busness, transportaton, telecommuncatons, an so on that are use as ntermeate nputs an subject to fferent barrers such as entry barrers an taxes per unt of prove servces. Markusen et al. (2005) use a moel wth monopolstc competton to show that foregn proucer servces coul actually prove substantal benefts to rval omestc frms. In partcular, rval frms beneft from the expanson of the servces nustry as a whole. Ths s because the omestc ownstream nustry purchases hgher qualty busness servces an expans as more foregn frms enter the omestc market. Hence, the optmal tarff s foun to be a subsy. In other wors, servces trae lberalzaton s foun to have a postve mpact not only on omestc consumers but also on rval frms an ownstream consumers. 4

13 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton However, partcularly n evelopng countres many backbone servces such as telecommuncatons, fnance an nsurance are characterze by olgopoly markets. Hoekman (2006), Mattoo an auve (2003) an others stress the mportance of market structure an regulaton for the outcome of servces trae lberalzaton. Furthermore, even though there are many trae moels that ncorporate market power, there are few that aress specal ssues relate to servces (Copelan, 2002). In contrast to stanar moels wth prefect competton, moels of mperfect competton coul prouce unexpecte results. It s suffcent to conser the case of a smple Cournot uopoly wth fferent margnal costs to obtan the counterntutve effects on welfare. In partcular, let us assume that margnal costs are constant an the margnal cost of the frst frm s hgher than the margnal cost of the secon frm. Then, uner certan parameters, the ecreasng of the margnal cost of the frst frm so that t stll remans hgher than the margnal cost of the secon frm brngs a negatve welfare effect. Ths happens because of the neffcency assocate wth the ncrease n the share of the less effcent frm. On the other han, n our moel, foregn frms are subject to hgher fxe costs compare to omestc frms an wll consequently ten to form more concentrate market structure. In aton to scrmnatory natonal treatment lmtatons such as specfc lcensng, foregn frms coul be expecte to have hgher fxe costs than omestc frms ue to aaptaton costs to operate n a new busness envronment, language barrers, etc. However, t s also plausble to assume that foregn frms are more effcent per unt of servce compare to local frms n evelopng countres. Even though the prouctvty fference n servces shoul be smaller than n manufacturng because of ther less traable nature, the counterntutve effects wth hgher markups n the foregn sector compare to the omestc wll be nval. In other wors, there wll be no smlar Cournot effect wth fxe costs as there s wth margnal costs. It coul be emonstrate that a possble negatve welfare effect n such case occurs rather because of the unstable ntal equlbrum. However, there s another source of neffcency assocate wth the msallocaton of monopoly markups between fferent sectors of the economy. In partcular, Epfan an Ganca (20) fn that trae lberalzaton that only apples to certan nustres coul lea to a negatve mpact on welfare when t rases markup heterogenety. A smlar argument coul be apple to our moel snce foregn an omestc servces are mperfect substtutes an coul be thought of as fferent sectors. Consequently, n orer to verfy f a smlar case s possble n our moel, we woul nee to have smaller fxe cost n the foregn nustry relatve to the omestc nustry because hgher markups are translate nto hgher fxe costs. Ths coul take place f the foregn frms possess sgnfcantly greater economes of scale an market expertse compare to omestc frms. Havng entry tax an the tax per unt of servce output woul not only enable us to moel the actual type of barrers that foregn frms face n evelopng markets, but also to fferentate between the mpact from fferent effcency of foregn frms n both varable an fxe costs. The purpose of ths chapter s therefore to bul upon the approach n Markusen et al. (2005) an construct an olgopoly moel wth conjectural output varatons n proucer servces n orer to stuy the mpact of servces trae lberalzaton on consumer welfare an omestc servce provers. Those wll epen on the mpact on foregn frms an the ownstream nustry that uses proucer servces as nputs. In partcular, the mpact of servces trae lberalzaton on omestc servces wll epen on the fference n effcency, economes of scale an fxe costs between foregn an omestc servce frms. Factors of proucton are supple nelastcally. Hence, changes n proucton of nustres wll efne the mpact on the 5

14 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton prces of factors of proucton. In partcular, mpact on sklle labor wll be rven by the relatve change n proucer servces whch use t more ntensvely than other sectors. On the other han, the mpact on unsklle labor woul be efne by the change n the more unsklle labor ntensve perfectly compettve sector. In orer to stuy the mpact of servces trae lberalzaton we wll nee to sentangle the relatve mportance an nteractons between such causal mechansms an effects as market crowng, preference for varety, pro compettve effect an so on. The stylze quanttatve moel s succnct an specfcally esgne to aress the gven problem. The mpact on the welfare of the host country woul epen on the magntue of those effects an coul among other thngs be negatve even when all the profts of the foregn frms accrue to the omestc representatve agent. On the other han, servces trae lberalzaton s expecte to brng hgher gans n the olgopoly moels compare to the monopolstc competton moels ue to the pro-compettve effect. The chapter s structure as follows: secton 2 presents a etale formulaton of the moel. The mpacts of some of the parameters of the moel on the markup are presente n secton 3. ecton 4 escrbes conseratons of the ata an the calbraton strategy use to obtan a benchmark replcaton of the moel. ecton 5 represents the numercal results of servces trae lberalzaton, scusses alternatve moel specfcatons an a numerare problem. ecton 6 conclues. The ervaton of the markup equaton an the program coe are gven n Appenx..2 Moel settng We assume that there are two sectors n the economy: Y enotes the sector that uses proucer servces as an ntermeate nput an Z s a perfectly compettve sector, whch coul also ncate all the other sectors of the economy. The consumers eman only fnal goos an consumer utlty s a Cobb ouglas composte of them: U Z Y (.2.) The Y sector s a CE (Constant elastcty of substtuton) composte of proucer servces an value ae: VA P Y ( Q Q ) (.2.2) Here QP enotes quantty of proucer servces an QVA enotes the value ae, θ>0 s a corresponng elastcty of substtuton. Analogous to Konan an Assche (2007) proucer servces are moele so that they postvely affect the value ae prouctvty when use as an ntermeate goo. The proucton structure of the Y sector s presente n Appenx A. Proucer servces, n turn, consttute an Armngton type CE functon of omestc an foregn servces: Q ( X ( ) XF ) (.2.3) P σ s an Armngton elastcty for servces. Whle there are no barrers to trae n fnal goos, trae n servces s subject to barrers an coul only be prove through commercal presence. 6

15 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton omestc (X) an foregn (XF) servces are aggregates of several varetes. Each frm prouces only one goo (varety) an competes n quanttes wth both omestc an foregn frms: X n ( x ) (.2.4) n j Corresponng ual prce nexes woul then be: n ( ) P ( p ) (.2.4') f f f f XF ( xf j ) (.2.5) n j f f j f PF ( pf ) (.2.5') Here n, n f are the number of omestc an foregn frms corresponngly. The proucton of each varety s subject to scale economes ue to fxe costs. It also employs omestc prmary factors of proucton: enotes sklle labor an L enotes unsklle labor or all the other factors of proucton: x L f f, n (.2.6) xf L j, n (.2.7) j j j f mlarly, the value ae of the Y sector an the perfectly compettve sector are the Cobb ouglas compostes of the prmary factors of proucton: VA Q L VA VA (.2.8) Z L (.2.9) I assume that the proucer servces sector s more sklle labor ntensve than the value ae an perfectly compettve sector an that foregn proucer servces are less sklle labor ntensve compare to omestc servces. In other wors, t s assume that:, (.2.0) f f VA In aton, I assume that foregn frms are more effcent relatve to omestc frms. There are few f no emprcal stues of comparatve effcency between omestc an foregn servce provers n evelopng countres (Whalley, 2004). The markup charge by proucer servces frms wll epen on the substtuton elastctes at fferent stages of proucton an the output conjectures of rval frms. In Appenx B, the perceve prce elastcty of eman s erve uner the assumptons that ) omestc frms make the same conjectures about the behavor of foregn frms as they o about the behavor of rval omestc frms, 2) there s symmetry n-between omestc frms an 3) all frms conjecture a constant value ae : H P ( ) ( ) / n (.2.) The markup for foregn frms coul be obtane analogously: f f f f f H P ( ) ( ) / n f (.2. ) f f f ln( xk ) Here s the conjectural elastcty of frm s output, s / n s a share of ln( x ) ln( ) the omestc frm n the total proucer servces nustry an Y ln( PY ) s the prce elastcty of eman n the ownstream nustry. f Ths assumpton follows from the moel structure. 7

16 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Then, the markup s obtane from the Lerner formula: ( px ) p MR p x p x x x p p x p p MC (.2.2) We coul see that the markup of frm ( ) s an nverse of the prce elastcty of eman. The markup equaton for foregn frms s calculate n the same way. Then, the Cournot f competton wll be a partcular case of ths settng wth 0 an 0. Another specfcaton woul be a cartel wth an f. n The aopte moelng approach assumes economes of scale at the frm level base on fxe costs. omestc an foregn frms are moele as representatve agents that receve markup revenues an pay fxe costs. In other wors, a zero proft conton at the omestc varety level wll look as follows: PX * n ( ) MC (.2.3) Here MC w v, w s a sklle wage an v s an unsklle wage. The prce per varety p, s etermne as follows: * p PX n Hence, the prce at the nustry level s ncreasng n the elastcty of substtuton between varetes an ecreasng n the number of frms. The total cost for the omestc nustry, uner zero proft conton, coul then be wrtten n the followng way: TC MC n X PX * X* (.2.4) mlarly, we coul wrte the zero proft conton for the foregn nustry: f f f n ( ) PXF TFX MC (.2.5) f If we put t nto wors, the foregn servces provers are subject to scrmnatory treatment n the form of per unt of output of servces tax (TFX) mpose only on foregn frms. I assume that costs are fxe n quanttes at the varety level (FC) an the secon term on the left han se of (.2.4) represents the total markup revenue of the omestc nustry. The latter s also the total fxe cost n values at the nustry level: PX * X * FC * P * n (.2.6) FC It wll not be qute fxe as t changes n terms of the prce of the fxe cost ( P n f FC ) an the number of frms. Ths s a very useful moelng approach that enables us to have a market power at the frm level an zero profts at the nustry level at the same tme. Moreover, I f assume that PFC MC an P FC MC. f The equvalent of the equaton (.2.6) for the foregn sector woul look as follows: The economes of scope coul also be ncorporate nto the settng by ang a fxe cost at an nustry level. 8

17 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton f f PXF * XF * FC * MC * n LT * n (.2.7) f f f Here LT enotes a lump sum or entry tax on foregn frms. It coul be thought of as a lcense fee that each foregn frm nees to pay n orer to access the omestc market. All the taxes are collecte by the omestc representatve consumer. A market clearng conton for omestc servces at an nustry level woul be: C( PX, PXF) X QP (.2.8) PX The left han se represents a eman for a omestc nustry goo by hepar s lemma an the rght han se s the quantty supple. Any conton coul be transferre to the varety level f we take nto account that: x n X (.2.9) In ths way, analogous to Markusen (2002), Markusen et al. (2005), the symmetry between varetes enables one to express the whole system at the nustry level only. Fnally, the market clearng conton for example for sklle labor coul be wrtten as follows n accorance to the structure of the moel: f PZ QVA MC f MC Z QVA n f ( xf FC ) n ( x FC ) (.2.20) w w w w.3. Partal ervatves of markup Evaluatng partal ervatves of the markup equaton wth respect to parameters coul alreay gve us some nsghts about the economc reasonng engrave nto the moel behavor. In partcular, the partal ervatves wth respect to prce elastcty of eman n ownstream nustry an the elastcty of substtuton between proucer servces an value ae are negatve uner conton (.3.): s (.3.) s n H P 2 ( ) 0 (.3.2) The hgher prce elastcty of eman n ownstream nustry wll translate nto a hgher responsveness of the latter to the changes n nput prces an consequently less market power n omestc proucer servces: H P 2 ( ) 0 (.3.3) mlarly, the market power of proucer servces wll eclne as the ownstream nustry gets more flexble n substtutng between ts prmary factors an servces nputs. The sgn of the partal ervatve wth respect to the Armngton type elastcty wll epen on the value of conjectural elastcty of frm output: 0 f 2 ( s )( ) / n (.3.4) 0 f Even f 0 the left han se shoul be negatve n most of the cases so that nequalty hols. 9

18 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton In partcular, uner relatvely more collusve conjectures of proucer servces frms about ther rvals output such that:, hgher elastcty of substtuton between omestc an foregn servces wll lea to a hgher markup of the omestc frms. On the other han, f servce proucers are relatvely more compettve n quanttes such that:, hgher Armngton type elastcty wll lea to a lower markup. Fgure.. The mpact of the elastcty of substtuton between varetes on markup epenng on the values of conjectural elastcty an the number of frms A smlar pattern coul be observe for the partal ervatve wth respect to the elastcty of substtuton between varetes: 2 2 ( n 3n ) (2 )( n ) (.3.5) ( ( n )) n For n 2.68, the relatvely collusve conjectures threshol wll be less than 0.5. In other wors, s suffcent for 0 an uner the Cournot competton ( 0) we 2 have 0 ; The shae area n Fgure. shows the regon n terms of n an where the mpact of the elastcty of substtuton on the markup of the omestc frm s negatve; The frms are better off havng more homogeneous goos uner collusve conjectures an the area where markup ecreases wth the elastcty of substtuton an expans wth the number of frms. Ths s also ntutve: more varetes woul ecrease the markup an ncrease competton. That woul make colluson less lkely to happen an consequently squeeze the regon where the mpact of the substtutablty on the markup s postve. The contons H P 0, an n are suffcent for the followng plausble propertes to hol: 0, 0 an 0. s n 0

19 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton amely, the markups of the omestc frms ncrease wth the omestc share an more collusve conjectures about the behavor of rval frms an rops n the number of omestc frms..4. ata n benchmark an moel calbraton The ata was structure so that the parameters fall nto reasonable ranges base on emprcal estmates an also reflect the assumptons mae wth respect to the structure of the moel. The mcro-consstency matrx (MCM) use for the benchmark calbraton coul be foun n Appenx. The ata from the note by the UCTA ecretatat UCTA (2003) reports that the share of proucer servces n GP s 20% on average for evelope countres an about 5% for least evelope countres. It s n the range of 7.5%-0% n evelopng countres. It coul be nferre from the MCM that the share of proucer servces of our moel s taken to be 0% of GP. The number of frms s set to be the same n foregn an omestc sectors an equals 0. In the base scenaro, I assume that the share of omestc nustry s equal to the share of foregn nustry. The foregn frms are subject to two types of scrmnatory barrers: per unt output tax an lump sum entry tax whch are taken to be equal to each other n value terms. The fnal goos are trae so that the goo from the sector wth mperfect competton s mporte an the goo from the perfectly compettve sector s exporte n the benchmark scenaro. I assume Cournot competton an ncreasng substtuton elastctes from the top level to the bottom n the proucton structure of the moel:,. There are several ways n whch a CGE moel wth mperfect competton coul be calbrate. I extraneously set the number of frms an the benchmark level of the markups an calbrate the bottom level elastcty of substtuton, resually (Gasorek et. al (992); Haalan an orman (992); Wllenbockel (994, 2004)). I use the value of.38 for foregn an.33 for omestc prce to margnal cost rato. Those values coul then easly be f transforme nto the markup value: 0.275, It shoul be PXF MC f even hgher uner ncreasng returns to scale (Chrstopoulou an Vermeulen, 2008). In partcular, the average prce to margnal cost ratos are estmate to be.56 for the Euro area an.38 for the U wth more varety across fferent sectors. However, takng hgher markup woul result n lower calbrate elastcty an unstable ntal equlbrum. Therefore the choce of the parameter values s n part etermne by stablty conseratons as t wll be further elaborate n the next secton. nce there s no consstent nformaton on the cost structure, the fxe cost rato s obtane resually from markup estmates. The moel s calbrate so that n the benchmark as many varables are equal to one as possble. Uner the assumpton of small open economy, the prce elastcty of eman n ownstream nustry shoul be equal to nfnty:. There are no precse estmates for the substtuton elastctes use n the moel. For example, there are many estmates of the Armngton substtuton elastcty between omestc an mporte goos (Mcanel an Barstrer, 2003). It s foun that stues wth hgher sectoral saggregaton obtan hgher Armangton elastctes. Consequently, statstcally sgnfcant elastctes are foun to be n the range of However, elastctes vary across sectors an t s ratonal to expect that elastctes of substtuton between servces are less than elastctes between goos. The f f

20 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton followng values of elastctes of substtuton are assume uner the base scenaro: 0.5, 2. Consequently, the calbrate bottom level elastctes are 5.4 an f 4.5. The factor ntenstes are assume to be n accorance wth assumptons mae on ther values: 0.3, 0.8, 0.75, 0.625, f VA Further conseratons on the stablty of the ntal equlbrum woul lea to a recalbraton of the parameters. If not specfe fferently, the exchange rate s taken as a numerare. The coe of the CGE moel s wrtten n the GAM syntax..5. Results of the numercal experments.5.. Results of the base scenaro The counterfactual polcy experments represent the free trae case when both taxes are lfte an cases wthout only output or entry tax. Table. reports the results of the counterfactual experments n percentages an results n levels are presente n Appenx C. Table.. Results of servces trae lberalzaton n percentages Percentage change from the benchmark Varables o entry o output Free trae tax tax Welfare (U) 2,7% 2,2% 3,4% Perfectly compettve sector (Z) -7,4% -5,6% -25,8% ownstream nustry (Y) 77,0% 78,4% 7,6% Proucer servces (Q P ) 85,2% 85,7% 32,7% Value ae (Q VA ) 69,5% 7,7% 04,4% omestc servces (X) 33,6% 46,5% 38,9% Foregn servces (XF) 45,2% 29,4% 250,6% umber of omestc frms ( ) 20,4% 28,5% 23,% umber of foregn frms ( f ) 80,3% 67,7% 237,4% et exports of Z (TEZ-TMZ) -22,0% -87,6% -309,0% et mports of Y (TMY-TEY) -95,2% -200,9% -30,2% Prce of the servces sector composte (P P ) -8,7% -7,7% -2,6% Prce of the value ae (P VA ) 8,9% 7,8% 3,2% Prce of the omestc servces (PX) 7,4% 3,9% 3,% Prce of the foregn servces (PXF) -20,7% -7,0% -28,8% Payments to the other factors of proucton (v) -7,4% -6,6% -0,7% Payments to sklle labor (w) 20,% 7,5% 30,6% Consumer ncome (CO) 2,8% 2,2% 3,5% Markup of omestc servce provers (MK) -4,4% -5,2% -5,2% Markup of foregn servce provers (MKF) -,6% -6,5% -2,4% hare of proucer servces (H P ) -4,4% -3,8% -6,4% hare of the omestc sector (sh ) -5,0% -,2% -22,8% hare of the foregn sector (sh f ) 5,0%,2% 22,8% Prce per foregn frm (pxf) 6,5% -3,8% 0,8% 2

21 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Output per foregn frm (xf) -34,6% 7,3% -26,9% Prce per omestc frm (px) 2,4% 0,3% 9,0% Output per omestc frm (x) 7,0% 8,8% 8,8% Welfare, measure as the Hcksan welfare nex, ncreases as a result of servces trae lberalzaton. In partcular, free trae brngs a 3.4% ncrease n welfare. It s also accompane by an expanson of ownstream nustry. The latter, n turn, stems from the equvalent expanson n proucer servces an the low prce of the foregn servces composte. Thus, a ecrease of ether output or entry tax wthn a reasonable range woul be welfare mprovng. The external sector s not comprehensvely moele an the rectons of change n trae flows are consstent wth change n proucton by the corresponng sector. In the servces sector: after taxes are lfte, more effcent foregn frms get a hgher share n nustry. Even though the omestc sector market share falls an as a result of that nustry prce ncreases, the ownstream nustry expanson leas to hgher eman for all of ts nputs an output of omestc servces ncreases. The expanson of both omestc an foregn servces s accompane by an ncrease n the number of varetes/frms. Ths hghlghts a procompettve effect that results n lower than benchmark markups. On the other han, the perfectly compettve sector (Z) contracts an payments to factors of proucton whch are use ntensvely n the proucton of that sector ecrease. Payments to sklle labor whch s use relatvely more ntensvely by ownstream nustry an by proucer servces ncrease after barrers on foregn provers are remove. The mpact on the prces of the factors of proucton coul potentally be fferent only f omestc servces ha a bgger share relatve to the rest of the economy an there was a greater fference n sklle labor ntenstes between omestc an foregn servces. The small open economy assumpton means that the prces of fnal goos are equal to worl prces an, as a consequence, they are vrtually fxe. Then, the varatons n prces of proucer servces an value ae woul be nhbte by the fxe prce n ownstream nustry. ubsequently, nustry level prces of omestc an foregn servces woul be constrane by the above mentone rgty of the composte prce of proucer servces. In other wors, the so calle rgty of those prces woul mply that a ecrease n the prce of foregn servces shoul be accompane by an equvalent ncrease n the prce of omestc servces. Moreover, because of the assumpton that, the change n the prce of the proucer servces composte wll be smaller than the change n the prce of the omestc or foregn servces nustry. Because of those restrctons on the changes n the prces of servces, apart from the rect prce effect, the results of the counterfactual experments wll be nfluence by other effects. In partcular, the xt-tgltz form of proucton functon efnes the preference for varety effect. Bascally t means that the total nustry output wll be value more by the consumers or the ownstream nustry than the total output of all frms combne because consumers attan atonal utlty from consumng varetes: XF xf * n, X x * n. mlarly, the nustry prce wll be lower wth more varetes. Therefore, ncrease n the number of frms pushes the prces at the nustry level ownwars f per varety prce s kept constant. However, n our moel, nustry level prces are constrane by the small open economy assumpton. One of the consequences of that stuaton coul be the occurrence of an unstable ntal equlbrum. Then, the system woul ajust so that the number of frms woul f 3

22 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton fall rather than ncrease after servces trae lberalzaton. It coul also happen that the prce per varety ncreases whle the nustry prce falls. Ths s exemplfe n the fference of the mpacts of two barrers for trae n servces. The results of the corresponng counterfactual experments are gven n columns an 2 of Table.. The barrers coul be unerstoo as a tax per frm or varety (LT) an a tax per unt of output of the frm (TFX). Consequently, a ecrease n the tax per frm facltates the entry of more foregn frms nto the market. Apparently the love-of-varety effect omnates over the constrane rect effect. It coul be seen from the ncreasng rather than ecreasng prce per foregn frm. The rse n nustry proucton s fully explane by the ncrease n the number of frms an the output per frm even falls. On the other han, smaller barrers per unt of output of each frm woul also lea to an ncrease n the number of varetes. Even though more varetes woul pressure the prce per foregn frm to ncrease, the rect prce effect omnates an t falls. In ths case, the preference for varetes effect constrans the expanson of foregn servces as the ncrease n the number of frms also consttutes a buren of hgher fxe costs. Consequently, there s more expanson n foregn servces as a result of the ncrease n proucton of each nvual frm rather than as a result of the rse n the number of frms. The relatve sgnfcance of the two nstruments of servces trae lberalzaton epens on the value of fxe costs, the elastctes of substtuton between varetes an sklle labor ntenstes. In partcular, removal of the entry tax compare to removal of the output tax has resulte n greater ncrease n the output of foregn servces an consequently greater share of proucer servces output. Therefore, both the share of the omestc sector an ts proucton are smaller uner no entry tax. That leas to less storton n the proporton of nputs of the ownstream nustry whch explans a somewhat hgher ncrease n the proucton of proucer servces uner no output tax. On the other han, the consumer welfare s hgher wth no entry tax because the representatve consumer receves more per unt of output tax revenues from the expanng foregn servces sector tablty of the ntal equlbrum It has been mentone that markups coul be expecte to have hgher values than assume n the benchmark. Let us take a somewhat hgher value of foregn markup: f 0.375, That woul correspon to a prce to margnal cost rato of.6 an lower than benchmark calbrate elastcty of substtuton 3 f. Uner lower elastctes of substtuton, the preference for varety effect ncreases. Ths mples more economes of versfcaton an shoul lea to an even greater expanson of ownstream nustry. However, as t coul be seen from Table.2, the economy lans at the other equlbrum pont wth a contracton n ownstream nustry an servces. In fact, the equlbrum escrbe n Table.2 s closer to the ntal equlbrum than the antcpate equlbrum wth substantal expanson n proucer servces. 4

23 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Table.2. Results of servces trae lberalzaton n percentages for hgher markup Percentage change from the benchmark Varables o entry tax o output tax Free trae Welfare (U) -,7% -,5% -2,2% Perfectly compettve sector (Z) 9,5% 7,3% 2,4% ownstream nustry (Y) -48,2% -35,5% -6,3% Proucer servces (Q P ) -49,3% -36,6% -62,4% omestc servces (X) -53,4% -37,5% -68,9% Foregn servces (XF) -45,% -35,7% -55,4% umber of omestc frms ( ) -38,5% -26,% -5,8% umber of foregn frms ( f ) -0,5% -28,8% -26,% Payments to the other factors of proucton (v) 4,4% 3,4% 5,8% Payments to sklle labor (w) -9,6% -7,4% -2,4% hare of the foregn sector (sh f ) 4,2% 0,8% 9,0% Prce per foregn frm (pxf) -4,9% -3,2% -6,3% Output per foregn frm (xf) -34,4% 6,3% -3,3% However, the counterntutve result presente n Table.2 occurs only because of the unstable ntal equlbrum wth hgher markup. A more etale analyss of equlbrum stablty an analytcal results on the mpact of the parameters on stablty wll be presente n the next chapter. The nstablty of the ntal equlbrum coul be emonstrate by changng the value of the exogenously etermne share of the prces of fnal goos ( PY PZ ) an observng the share of the corresponng outputs (Y Z ).,02,0 PY/PZ Intal o Output tax,00 E LT E TFX E 0 o Entry tax 0,99 0,0 0,5,0,5 Y/Z Fgure.2. The share of the outputs of fnal goos for fferent values of the prce share (An unstable ntal equlbrum case) The ownwar slopng curve at the pont of ntal equlbrum E 0 s a sgn of an unstable equlbrum (Fgure.2). Therefore, removal of the barrers of trae n servces results n the leftwar shft of the ntal curve an the assocate counterntutve result. We enote the 5

24 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton * * PY share of the worl prces of two goos as P an the relatve supply prce of the two * PZ goos obtane from the proucton structure of the moel as the functon of taxes an the Y rato of outputs: P ( t, ). Then, the Marshallan output ajustment stablty conton v coul Z be wrtten as follows: 0 0 * Y Y Y P P t (, ) * 0 (.5.) Z Z Z 0 0 Y Here t { TFX, LT} an are the values of the barrers an the output rato n the Z benchmark. In other wors, the followng shoul hol: P * 0 0 Y P ( t, ) 0 (.5.2) Z Y t an are the corresponng values after one of the two barrers for the servces trae are Z remove. The correct result wth substantal expanson n ownstream nustry s presente n Table.3. Table.3. Results of servces trae lberalzaton n percentages Percentage change from the benchmark Varables o entry tax o output tax Free trae Welfare (U) 9,2% 7,2% 0,2% Perfectly compettve sector (Z) -40,3% -34,7% -49,5% ownstream nustry (Y) 204,8% 83,2% 253,9% Proucer servces (Q P ) 237,9% 209,2% 30,8% omestc servces (X) 48,2% 6,% 38,5% Foregn servces (XF) 504,8% 405,2% 702,0% umber of omestc frms ( ) 27,9% 35,9% 2,5% umber of foregn frms ( f ) 295,2% 67,8% 347,4% Payments to the other factors of proucton (v) -5,6% -3,6% -8,6% Payments to sklle labor (w) 49,% 4,2% 62,0% hare of the foregn sector (sh f ) 33,8% 27,8% 4,2% Prce per foregn frm (pxf) 20,8% 7,3% 6,0% Output per foregn frm (xf) -25,0% 2,5% -5,6% 0 Y Y Hence, f Z Z 0, for the Marshallan output ajustment conton to hol, t s Y P ( t, ) Z suffcent f the followng conton hols: 0 (.5.3). t 0 v A more comprehensve scusson of stablty contons s gven n the next chapter. 6

25 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Moreover, f we take nto account that ual to (.2.2) PY ( P P ), ual to (.2.8) VA VA PVA w v an ual to (.2.9) PZ w v, the mpact of taxes on the relatve supply prce wll be manly etermne by the mpact on the composte prce of proucer servces: 0 Y PP ( t, ) Z 0 (.5.4) t nce proucton of fnal goos s assume to be at the benchmark, the mpact of the barrers on proucer servces wll be efne only by the rect mpact on foregn proucers. Consequently, equatons (.2.5) an (.2.7) coul be wrtten n the followng way uner the benchmark: 0 f 0 f 0 0 TFX MC FC f * MC * n f LT * n f PXF (.5.5) an PXF (.5.6) 0 0 f 0 0 f XF n f ( ) f Then, from (.5.5) an (.5.6) t follows that: 0 0 Y Y PXF ( TFX, ) PXF ( LT, ) Z Z 0 (.5.7), 0 (.5.8) TFX LT Those nequaltes mean that the fall n relatve output of the ownstream nustry: 0 Y Y The stuaton wth 0coul only occur when the LT an TFX tax barrers are Z Z ncreasng. The graph of the relatve output of ownstream nustry corresponng to the stable equlbrum of Table.3 s presente n Fgure.3. VA P,02 PY/PZ,006 Intal o Output tax E 0 E TF E LT o Entry tax 0, Y/Z Fgure.3. The share of the outputs of fnal goos for fferent values of the prce share (A stable ntal equlbrum case).5.3. umerare problem General equlbrum moels wth mperfect competton an scale economes coul be prone to the numerare problem. In other wors, as emonstrate by Gabszewcz an Val (972) an Mercener (995) among others, the choce of the numerare of the general equlbrum 7

26 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton moel may brng sgnfcantly fferent results. Ths shoul not be the case an all of the values shoul just be monotonc transformatons of each other uner fferent varables as a numerare. In general equlbrum ths coul be emonstrate by the Walras law accorng to whch any one equaton of the moel always hols as long as the remanng equatons o hol. In other wors, any of the prces coul be fxe an taken as a numerare an ths shoul not change the results substantally. In the frst column of Table.4 the results of the full servces trae lberalzaton wth the prce of proucer servces as a numerare are presente. Table.4. Results of servces trae lberalzaton n percentages for fferent numerare values Percentage change from the benchmark after free trae Varable/umerare P P P P P VA an 0. Welfare (U) -2,3% 0,4% 3,5% Perfectly compettve sector (Z) 4,5% -3,2% -26,% ownstream nustry (Y) -69,8% 4,7% 22,2% Proucer servces (Q P ) -74,0% 3,2% 37,9% Value ae (Q VA ) -68,7% 0,9% 03,% omestc servces (X) -82,4% -28,4% 4,7% Foregn servces (XF) -63,9% 64,5% 258,8% umber of omestc frms ( ) -65,9% -20,8% 25,0% umber of foregn frms ( f ) -28,7% 96,3% 243,4% Prce of the servces sector composte (P P ) 0,0% 0,0% -23,% Prce of the value ae (P VA ) -3,3% 4,9% 0,0% Prce of the omestc servces (PX) 2,6% 25,8% -0,4% Prce of the foregn servces (PXF) -5,% -7,0% -37,4% Payments to the other factors of proucton (v) -0,3% 5,9% -2,4% Payments to sklle labor (w) -20,2% 20,6% 5,5% Consumer ncome (CO) -8,9% 7,7% -8,7% Markup of omestc servce provers (MK) 33,2% 3,6% -5,2% Markup of foregn servce provers (MKF) 7,3% -8,4% -2,4% hare of proucer servces (H P ) 9,2% -,0% -7,8% hare of the omestc sector (sh ) -7,8% -20,6% -22,8% hare of the foregn sector (sh f ) 7,8% 20,6% 22,8% Prce per foregn frm (pxf) -9,2% 2,2% 42,3% Output per foregn frm (xf) -44,2% -30,8% -26,9% Exchange rate (e) -6,7% 7,3% -,9% Inee, the result s substantally fferent from the result wth the benchmark numerare of exchange rate presente n the thr column of Table.. However, the equlbrum outcome s smlar to the result presente n Table.2. It coul be shown that ths result s also unstable an generates the same pattern of ratos of prces an outputs of fnal goos as presente n Fgure.2. Therefore, the ntal equlbrum shoul be recalbrate to a stable one. Choosng the prce of proucer servces (P P ) as a numerare woul lmt the rect effect of prce 8

27 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton ajustment even more than the assumpton of a small open economy. Ths s because the latter only fxes the prces of fnal goos. One of the mechansms that coul mprove the stablty of the ntal equlbrum s the market crowng n sklle labor effect. Ths effect wll be stronger the hgher s the fference between, f, VA an an the lower s the value of f. v In partcular, the secon column of Table.4 presents the same result but wth a new value of unsklle labor ntensty: 0.. Ths tme, the result follows the pattern of Tables. an.3 an the equlbrum outcome s stable wth the same slope as n Fgure.3. The mpact s just a monotonc transformaton of the ntal result. Consequently, n ths moel, the numerare problem coul only occur because of the nstablty of the resultant ntal equlbrum. mlarly to the prce of proucer servces, takng PXF an PX as a numerare also unermnes the rect effect even more an prouces the same result n terms of the rectons of change of the varables as the frst column of Table.4. Takng the prce of unsklle labor as a numerare also leas to the same result as wth PXF, PX an P P. However, t unermnes the market crowng effect. All of those cases coul be recalbrate to a stable result equvalent to column two of Table.4 by takng 0.. On the other han, takng the prce of the value ae of the ownstream nustry (P VA ) or the prce of the sklle labor as a numerare represent just the monotonc transformatons of the ntal result as t coul be seen from the results presente n the thr column of Table.4. klle labor represents a small share of the total value ae an therefore takng the prce of sklle labor as a numerare s nsuffcent to obtan an unstable result. There s another aspect assocate wth the numerare problem n the general equlbrum moels wth olgopoly an monopoly market structure. amely, f the monopoly or olgopoly frm s large relatve to the market sze of the gven economy, ts prcng strategy coul have repercussons on the prces of other goos an consumer ncome. Consequently, f such feeback effects are taken nto account by the respectve olgopoly/monopoly frm, the choce of the numerare woul be mportant n etermnng outcomes. However, f the share of the olgopoly frm s small relatve to the rest of the economy, whch shoul be the case for most of the economes an frms, the results obtane uner full cognton of the feeback effects an wthout them woul not be much fferent from each other. Ths problem has been stue numercally usng a prototype moel n Wllenbockel (2005). It was foun that f the share of the frm s less than 0% of the economy, the results wth an wthout the full cognton of the feeback effects wll be approxmately the same. In our moel, the share of the whole proucer servces nustry s equal to 0% of the GP. Therefore, ths type of the problem s not relevant to our moel an there s no nee to calculate olgopoly profts wth full cognton of the feeback effects Other specfcatons of the moel Let us ntrouce some other changes to the benchmark settng. In orer to stuy the mpact of market structure on moel behavor, we coul change the conjectures of the frms about the proucton strategy of ther rvals. In partcular, let us assume perfect colluson such that: n an f n. The elastcty of substtuton between proucer servces an the value f ae an the elastcty of substtuton between foregn varetes shoul be recalbrate n v A more etale scusson of the market crowng effect as well as the analytcal ervaton of the parameters that coul strengthen t wll be gven n the next chapter. 9

28 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton orer to preserve the assumptons on the parameters:.5, 3.02, 3. The results of the free trae n servces scenaro s presente n the frst column of Table.5. The results are of a somewhat greater scale ue to the hgher value of the elastcty of substtuton between value ae an proucer servces. Uner collusve conjectures, the reactons of the frms are coornate to realze the hghest nustry proft. Even though the conjectures of the frms are entrely fferent from Cournot, the mpact of servces trae lberalzaton has not been fferent for most of the varables. The man fference s n the way an expanson of the ownstream nustry an proucer servces s realze. amely, the number of foregn frms oes not ncrease as much as uner the Cournot conjectures an the number of omestc frms even falls. Ths behavor of the moel shoul be expecte n the less compettve collusve settng. The preference for varety effect constrans sgnfcant contracton n the number of frms. On the other han, output per frm substantally ncreases after servces trae lberalzaton an the prce per varety falls or ncreases much less than before. However, the mpact on welfare, the output of servces an fnal goos as well as the prces of the factors of proucton s the same as t was uner the Cournot competton. Table.5. Results of servces trae lberalzaton n percentages Percentage change from the benchmark f Varables Free trae an Collusve conjectures Welfare (U) 2,9% Perfectly compettve sector (Z) -23,4% ownstream nustry (Y) 35,8% Proucer servces (Q P ) 77,2% Value ae (Q VA ) 98,8% omestc servces (X) 56,6% Foregn servces (XF) 332,0% umber of omestc frms ( ) -28,9% umber of foregn frms ( f ) 28,0% et exports of Z (TEZ-TMZ) -277,2% et mports of Y (TMY-TEY) -352,% Prce of the servces sector composte (P P ) -0,3% Prce of the value ae (P VA ) 2,0% Prce of the omestc servces (PX) 9,3% Prce of the foregn servces (PXF) -28,2% Payments to the other factors of proucton (v) -9,7% Payments to sklle labor (w) 27,4% Consumer ncome (CO) 2,9% Markup of omestc servce provers -54,8% o entry tax an Lower foregn markup o output tax Lower omestc markup an 0. 0,7% -0,3% -7,7% -0,5% 6,6% 2,9% 9,% 3,4% 4,3% 2,4% -5,3% -20,2% 46,3% 30,0% -4,% -,2% 27,2%,9% -88,8% -2,% -4,9% -8,9% -4,% -,0% 4,%,0% 7,6% 2,7% -3,4% -,7% -3,7% -0,2% 9,%,7% 0,7% -0,3% 0,4% 0,0% 20

29 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton (MK) Markup of foregn servce provers (MKF) -73,8% hare of proucer servces (H P ) 5,6% hare of the omestc sector (sh ) -24,8% hare of the foregn sector (sh f ) 24,8% Prce per foregn frm (pxf) 3,0% Output per foregn frm (xf) 96,9% Prce per omestc frm (px) -5,7% Output per omestc frm (x) 64,5% -24,6% -2,3% -2,0% -0,4% -0,8% -2,2% 0,8% 2,2% 5,4%,3% -60,0% 4,7% -,0% -0,% 0,0% 0,0% We have alreay seen that lower than benchmark elastcty of substtuton between varetes coul lea to an unstable ntal equlbrum. Let us now conser the case of hgher than benchmark value of that elastcty as a result of a smaller calbrate markup an fxe cost. The outcome of removng an entry tax s presente n the secon column of Table.5. The nteracton between the two effects coul explan the fference wth the base scenaro. On one han, lower markup shoul lea to more competton an stronger expanson of the servces sector. Moreover, lower foregn fxe cost woul mean the entrance of more effcent foregn frms nto the market. Inee, f we compare our result wth the frst column of Table., foregn servces acqure greater share, more foregn frms enter the market an the markup falls after the removal of an entry tax. On the other han, hgher elastcty of substtuton between varetes leas to less economes of versfcaton an a smaller fxe cost also mples smaller economes of scale effect. Ths explans why the expanson n servces an ownstream nustry s smaller than before. fferently from the base scenaro n Table., the omestc sector contracts. Therefore, the expanson of ownstream nustry wll be smaller the lower s the value of the Armngton elastcty of substtuton between omestc an foregn servces. Fnally, let us assume that n contrast to the effcent foregn servces sector, the omestc servces face large fxe costs an the corresponng elastcty of substtuton s low. Ths experment s one n orer to test f t s possble to obtan a stuaton such that the contracton of the omestc sector s so strong that t leas to the contracton of the ownstream nustry. Inee, such result coul be attane wth 9, 2.2. Even though foregn servces expan, accorng to conton (.5.) the result s unstable. Thus, takng hgher markup an lower elastcty of substtuton n the omestc nustry coul also lea to an unstable result. However, f we ajust parameters affectng the market crowng effect, take 0., a stable equlbrum outcome wth expanng foregn servces an ownstream nustry but contractng welfare coul be obtane as shown n the thr column of Table.5. The ncrease n the proucton of the ownstream nustry s moest because of the relatvely low substtutablty between foregn an omestc servces an the hgh fference between them. In other wors, f there s a sgnfcant msallocaton between omestc an foregn servces n favor of foregn servces, the omestc economes of scale are constrane an servces trae lberalzaton only makes that msallocaton an seconomes of scale stronger. That brngs a negatve welfare effect. Consequently, servces trae lberalzaton coul have a negatve mpact not only on the proucton of the omestc sector but also on welfare. f 2

30 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Ths outcome coul be expecte to occur for example n telecommuncatons or the transportaton servces sectors. The omestc ncumbents may have property rghts over nfrastructure an there coul be some local characterstcs that cannot be easly acqure or local legslaton that cannot be easly ajuste to by foregn frms. That shoul lea to a relatvely small estmate of the elastcty of substtuton between omestc an foregn servces an relatvely hgher economes of scale n the omestc nustry. Those are the reasons that lea to the result presente n Table.5. On the other han, n the fnancal an busness servces there shoul be less of those kns of barrers an therefore the estmate of the Armngton substtuton elastcty coul be expecte to be hgher. Moreover, the fferences n prouctvty an the entry costs coul be smoothe by relatve spllover effects. Ths coul lea to an expanson of omestc servces as well as foregn servces an follow the base scenaro of our moel..6. Conclung remarks The moel constructe n ths chapter has emonstrate several key results mportant for polcy analyss an further emprcal stues. In partcular, t has been foun that: Trae lberalzaton n proucer servces leas to an expanson of ownstream nustry that uses them as nputs. A very hgh level of markup n both omestc an foregn servces coul lea to an unstable equlbrum. In other wors, conseratons about the nstablty of the ntal equlbrum coul mpose atonal constrants on the value of the moel parameters. The conjectures of frms about the proucton strategy of ther rvals have lttle mpact on moel results. On the other han, changes n the elastctes of substtuton coul have more profoun effects on the moel results compare to conjectures of the frms. Hgh fference n markups between omestc an foregn frms coul lea to an nsgnfcant expanson n ownstream nustry an a negatve welfare effect. Moreover, t was foun that the two common types of barrers on the operatons of foregn frms woul have a smlar mpact on the recton of change of the key varables of the moel. However, they efne two fferent patterns of expanson of the foregn servces nustry. The analytcal results of the relatonshps between parameter values an the markup have also been presente. The expanson of ownstream nustry as a result of the lberalzaton of trae n proucer servces coul also be explane as follows: If trae protecton of ntermeates nclung servce ntermeates s hgher than trae protecton of ownstream nustry, the effectve rate of protecton (ERP) n ownstream nustry s negatve. That means that ownstream nustry value ae s effectvely taxe not protecte. Therefore, cuttng trae barrers on proucer servces turns ERP for ownstream nustry to be postve an that leas to an expanson of the latter. It shoul be ae that the magntues of the numercal results shoul be accepte wth ue cauton because of the stylze nature of the unerlyng moel. In other wors, t s rather narrowly focuse an emphaszes partcular mechansms only. On the other han, a stylze quanttatve moel coul serve as a useful tool to sentangle the varous effects present n the more complcate apple moels. Those latter moels are frequently vewe as black boxes 22

31 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton because t coul be ffcult to explan whch mechansms an effects are responsble for the results (evarajan, Robnson, 2005). Both apple an stylze moels shoul be combne to be use effectvely n polcy analyss. Improvements n the ata on servces coul sgnfcantly enhance the power of the analyss. In the moel settng presente n ths chapter, nvual frm effcency an fxe cost share epen postvely on each other. The fferences n fxe cost share between foregn an omestc frms appear to be crtcal for the negatve welfare effect obtane from the polcy experments n the numercal moel. However, there s a lack of ata on the comparatve effcency of foregn versus omestc frms n servces. Bblography Chrstopoulou, R., Vermeulen, P., (2008) "Markups n the euro area an the U over the pero a comparson of 50 sectors" Workng Paper eres 856, European Central Bank. Copelan, B., (2002) Benefts an costs of trae an nvestment lberalzaton n servces: Implcatons from trae theory n Curts, J. M. an. C. Curak, es. Trae Polcy Research epartment of Foregn Affars an Internatonal Trae, Canaa, pp evarajan,., Robnson,., (2005) The Influence of Computable General Equlbrum Moel on Polcy n Kehoe. T, rnvasan, T.. an Whalley, John (es) Fronters n Apple General Equlbrum Moellng. Cambrge Unversty Press, pp Epfan, P., Ganca, G., (20) Trae, markup heterogenety an msallocatons, Journal of Internatonal Economcs (83), pp. -3. Gabszewcz, J.J., an Val, J.-P., (972) Olgopoly A la Cournot n a General Equlbrum Analyss, Journal of Economc Theory (4), pp Gasorek, M., mth, A., Venables, A., (2002) The accesson of the UK to the EC: a welfare analyss. J. Common Mark. tu. 40, pp Haalan, J., orman, V., (992) Global proucton effects of European ntegraton In: Wnters, L.A. (es), Trae Flows an Trae Polcy after 992, Cambrge Unversty Press, Cambrge, pp Hoekman, Bernar. (2006). Lberalzaton of Trae n ervces: A urvey, Worl Bank workng paper. Konan,., Assche, A., (2007). Regulaton, Market tructure an ervce Trae Lberalzaton, Economc Moelng 24 (2007), pp Markusen, J., (989) Trae n Proucer ervces an n Other pecalze Intermeate Inputs, Amercan Economc Revew (79), pp Markusen, J., (2002) General Equlbrum Moelng usng GAM an MP/GE. Bouler. Markusen, J., Rutherfor, T., an Tarr,., (2005) Trae an rect nvestment n proucer servces an the omestc market for expertse, Canaan Journal of Economcs 38 (2005), pp Mattoo, A., Rathnran, R., an ubramanan, A., (2006) Measurng ervces Trae lberalzaton an ts Impacts on Economc Growth: An Illustraton, Journal of Economc Integraton 2 (2006), pp Mattoo, A., auve, P., (2003) omestc Regulaton an ervces Trae Lberalzaton, Oxfor Unversty Press: Washngton,.C. Mcanel, C., an Balstrer, E., (2003) A Revew of Armngton Trae ubsttuton Elastctes, Économe Internatonale, 94-95, pp

32 Chapter. Trae n Intermeate Proucer ervces uner Imperfect Competton Mercener, J., (995) onunqueness of olutons n Apple General Equlbrum Moels wth cale Economes an Imperfect Competton, Economc Theory (6), pp Rutherfor, T., Tarr,., an hepotylo, O., (2006) The Impact of WTO Accesson an the A: The Importance of Lberalzaton of Barrers Aganst FI n ervces for Growth an Poverty Reucton, n Thomas Hertel an L. Alan Wnters (es.), Poverty an the WTO (Palgrave an Worl Bank), pp e ants, R., (2002) A computable general equlbrum moel for open economes wth mperfect competton an prouct fferentaton. J. Econ. Integraton (7), pp UCTA (2003). Assessment of Trae n ervces an evelopment Gans, Hanbook of tatstcs. raft note by the UCTA secretarat. Whalley, J., (2004) Assessng the Benefts to evelopng Countres of Lberalzaton n ervces Trae, The Worl Economy 27(8), pp Wllenbockel,., (994) Apple General Equlbrum Moellng: Imperfect Competton an European Integraton John Wley, Chchester. Wllenbockel,., (2004) pecfcaton Choce an Robustness n CGE Trae Polcy Analyss wth Imperfect Competton, Economc Moelng (2), pp Wllenbockel,., (2005) The Prce ormalzaton Problem n General Equlbrum Moels wth Olgopoly Power: An Attempt at Perspectve, Mlesex Unversty Busness chool scusson Paper eres: Economcs. 24

33 Chapter. Appences Appenx A: Fgure.A. The proucton structure of the Y nustry Y CE: θ VA CE: P CE:σ L X XF CE: σ f CE: σ x x n xf xf n Appenx B: ervaton of the markup equaton nce the ownstream nustry s assume to be perfectly compettve, we have: Y Y QP X p Py Py ( A) x QP X x The partal ervatves obtane from the above ecomposton coul easly be calculate as follows: Y QP QP Y (2 A) QP P X X (3 A) X X x A x If we plug (2A), (3A), (4A) nto (A), we coul obtan the followng expresson: Y QP y Q P X L L L L p P X x (5 A) In orer to rewrte (5A) n terms of the relatve changes, I aopt the so calle hat notaton that s routnely use n trae textbooks: ˆ Y Yˆ Y log( Y ). For example n that case, s the Y P prce elastcty of eman n the Y sector. After mposng the neee transformatons on (5A): ˆ ˆ p Py Y Q P QP X X x (6 A) (4 ) ˆy 25

34 Chapter. Appences If we ve (6A) by x, we obtan the nverse of the perceve elastcty of eman : p ˆ Y Q P X x x x x (7 A) Yˆ xk Here s the aggregate prce elastcty an s a conjectural elastcty of Pˆy x output. Pˆ ˆ ˆ ˆ y Py Y Y I coul use the followng transformaton to further smplfy (7A):. Yˆ x x x The expressons for X an P are easly transforme nto hat notaton: x k X x k. k x j j Here I enote sh k j x k x j as a share of the frm n the whole omestc nustry an X ( ) XF Q X XF s X s XF. P f X ( ) XF X ( ) XF Consequently, s s a share of the omestc sector: s X X ( ) XF The followng conton shoul hol uner the symmetry n-between the omestc varetes: shk. n ubsequently, assumng that ) omestc frms make the same conjectures about the behavor of foregn frms an 2) symmetry n-between omestc frms the followng expressons coul be erve: x X k xk x k * sh ( ) / (8 ) k shk sh n A k x k k x x x j j QP X XF s ( s ) s ( s ) n f ( ) (9 A) x x x n n n f Here s / n s the share of the omestc frm n the total proucer servces nustry.. 26

35 Chapter. Appences The proucton of the ownstream nustry coul be wrtten n hat notaton as follows: Yˆ Q H Q H (0 A) VA VA P P H P Q QP VA QP Yˆ QP QVA Then, H P ( A). By movng from (0A) to (A) we assume that 0. In x x x other wors, snce value ae s gven as a composte of the prmary factors, t s assume that servce frms assume that ther proucton ecsons wll have no mpact on the value ae. If we plug (8A), (9A), an (A) nto (7A), we fnally obtan the neee markup equaton for omestc frms: H P ( ) ( ) / n (.2.) Appenx C: Results of the senstvty analyss Table.C. Results of servces trae lberalzaton n levels Change n levels from the benchmark Varables Benchmark o entry tax o output tax Free trae Welfare (U),027,022,034 Perfectly compettve sector (Z) 0,826 0,844 0,742 ownstream nustry (Y),77,784 2,76 Proucer servces (Q P ),852,857 2,327 Value ae (Q VA ),695,77 2,044 omestc servces (X),336,465,389 Foregn servces (XF) 2,452 2,294 3,506 umber of omestc frms ( ) 0 2,04 2,846 2,307 umber of foregn frms ( f ) 0 28,03 6,768 33,74 et exports of Z (TEZ-TMZ) -,2-0,876-2,09 et mports of Y (TMY-TEY) -0,952 -,009-2,02 Prce of the servces sector composte 0,93 0,923 0,874 (P P ) Prce of the value ae (P VA ),089,078,32 Prce of the omestc servces (PX),074,039,3 Prce of the foregn servces (PXF) 0,793 0,83 0,72 Payments to the other factors of 0,926 0,934 0,893 proucton (v) Payments to sklle labor (w),20,75,306 Consumer ncome (CO) , , ,672 Markup of omestc servce provers (MK) 0,25 0,239 0,237 0,237 27

36 Chapter. Appences Markup of foregn servce provers 0,275 0,243 0,257 0,24 (MKF) hare of proucer servces (H P ) 0,5 0,478 0,48 0,468 hare of the omestc sector (sh ) 0,5 0,425 0,444 0,386 hare of the foregn sector (sh f ) 0,5 0,575 0,556 0,64 Prce per foregn frm (pxf),93 2,056,858,946 Output per foregn frm (xf) 0,052 0,034 0,06 0,038 Prce per omestc frm (px),743,959,923 2,074 Output per omestc frm (x) 0,057 0,06 0,062 0,062 28

37 Chapter. Appences Appenx : Table.. Mcro-consstency matrx (MCM) Proucton sectors Markets Z Y Q P X XF f TEZ TMY U CO E EF PZ 3200,0-300,0-2900,0 PY 800,0 300,0-00,0 P P -400,0 400,0 PX -200,0 200,0 PXF -200,0 200,0 PF 50,0-50,0 PFF 50,0-50,0 PU 4000,0-4000,0 v -2240,0-50,0-30,0-25,0-0,0-2,5 2467,5 w -960,0-250,0-20,0-75,0-40,0-37,5 482,5 e 300,0-300,0 Markup Revenue () -50,0 50,0 Markup Revenue (F) -50,0 50,0 Output Tax -25,0 25,0 Entry Tax -25,0 25,0 Here P P s a prce of the proucer servces composte, PF an PFF are prces of the fxe costs at a varety level for omestc an foregn frms corresponngly, PU- consumer welfare prce nex, TEZ-export of goo Z, TMY-mport of goo Y. w an v are returns for sklle labor an other factors of proucton corresponngly. e s an exchange rate. E an EF represent the agents that receve the markup revenue from omestc an foregn frms. All the numbers are gven n values. The row an column sums are equal to zero. 29

38 Chapter. Appences Appenx E: Program coe $TITLE Moel of ntermeate proucer servces lberalzaton uner olgopoly wth output conjectures * For gven number of frms an estmate markup, the elastcty of substtuton s calbrate * Government has two polcy tools: lump sum tax an output tax PARAMETER THETA Elastcty of substtuton between value ae an proucer servces, IGMA omestc frm eman elastcty, IGMAF FI frm eman elastcty, IGMA Armngton elastcty, OMEGA Elastcty of eman for goo Y, TFX Output tax on foregn servces prover, ELTA The share parameter of the Armngton functon, v Conjectural elastcty of output for omestc frms, vf Conjectural elastcty of output for foregn frms, LT Value of the lump sum tax, C Calbraton parameter, FC Calbraton parameter, C Calbraton parameter, FC Calbraton parameter, B Calbraton parameter, PEZ Export prce of goo Z, PMZ Import prce of goo Z, PEY Export prce of goo Y, PMY Import prce of goo Y, ALPHA The share of goo X n the consumer expenture, L_EOW Total Enowment of klle labor, _EOW Total Enowment of Unsklle labor, GAMMA klle labor ntensty of the omestc servces, GAMMAF klle labor ntensty of the foregn servces, BETA klle labor ntensty of Z sector, GAMMAVA klle labor ntensty of the value ae composte, FC Fxe cost of the omestc frm, FCF Fxe cost of the foregn frm, pxf0 Frm level prce of the foregn servces, xf0 Frm level output of the foregn servces, px0 Frm level prce of the omestc servces, x0 Frm level output of the omestc servces ; FC=5; FCF=3; ALPHA=0.725; L_EOW= ; _EOW= 482.5; GAMMA=0.8; GAMMAF=0.75; BETA=0.3; GAMMAVA=0.625; THETA = 0.5; IGMAF =4.5; IGMA =5.4286; IGMA =2; ELTA =0.5; v = 0; vf = v; TFX=0.25; LT=2.5; C= 0**(/(-IGMA)); 30

39 Chapter. Appences FC= 0**(/(-IGMAF)); PEZ=; PMZ= 000/999; PEY=999/000; PMY=; B= 2**(/(THETA-)); POITIVE VARIABLE W! Consumpton (statc welfare) level Z! Actvty level for sector Z Y! Actvty level for sector Y P! Intermeate proucer servces VA! Value ae X! Actvty level for sector X omestc servces XF! Actvty level for sector XF foregn (ME) servces! umber of omestc servce frms F! umber of foregn (ME) servce frms TMZ! Actvty level for mports of Z TEZ! Actvty level for exports of Z TMY! Actvty level for mports of Y TEY! Actvty level for mports of Y PW! Prce nex for consumpton PZ! Prce nex for commoty Z PY! Prce nex for commoty Y PP! Prce nex for proucer servces PVA! Prce nex for the value ae PX! Prce nex for omestc commoty PF! Prce of fxe costs for omestc servces PXF! Prce nex for commoty Z ME PFF! Prce of fxe costs for ME servces RL! Return to unsklle labor L (composte) R! Return to sklle labor (sklle labor) PFX! Prce nex for foregn exchange CO! Income level for consumer CO ETRE! ummy agent for omestc servces (receves markups) ETREF! ummy agent for foregn servces (receves markups) MK! Markup by omestc frms MKF! Markup by foregn frms HP! hare of the proucer servces sh! hare of the omestc nustry n the proucer servces shf! hare of the foregn frms n the proucer servces ; EQUATIO PrceW Zero proft conton for welfare PrceZ Zero proft conton for Z PrceY Zero proft conton for Y PrceVA Zero proft conton for VA PrceP Zero proft conton for Proucer ervces PrceX MR=MC conton for omestc frms PrceXF MR=MC conton for foregn frms PrcePF Zero proft conton for omestc fxe costs PrcePFF Zero proft conton for foregn fxe costs E_TEY Zero proft conton for exports of Y E_TMZ Zero proft conton for mports of Z E_TEZ Zero proft conton for exports of Z E_TMY Zero proft conton for mports of Y W eman for W Z eman for Z Y eman for Y 3

40 Chapter. Appences X eman for X XF eman for XF PFX Trae balance equaton VA eman for value ae P eman for proucer servces RL eman for unsklle labor R eman for sklle labor Zero proft conton for omestc frms F Zero proft conton for foregn frms ICO Income balance of the representatve consumer IETRE Income balance of the omestc frms IETREF Income balance of the foregn frms harep Proucer servces share balance haresh omestc share balance hareshf Foregn share balance Markup omestc markup equaton MarkupF Foregn markup equaton ; * Zero proft contons: PrceW.. PZ**ALPHA*PY**(-ALPHA)=G=PW; PrceZ.. R**BETA*RL**(-BETA)=G=PZ; PrceY.. B*(PP**(-THETA)+PVA**(-THETA))**(/(-THETA))=G=PY; PrceVA.. R**GAMMAVA*RL**(-GAMMAVA)=G=PVA; PrceP.. (ELTA*(PX)**(-IGMA)+(-ELTA)*(PXF)**(-IGMA))**(/(-IGMA))=G=PP; PrceX.. R**GAMMA*RL**(-GAMMA)=G=**(/(IGMA-))*PX*(-MK)*C ; PrceXF.. R**GAMMAF*RL**(-GAMMAF)+TFX=G=F**(/(IGMAF-))*PXF*(-MKF)*FC; PrcePF.. R**GAMMA*RL**(-GAMMA)=G= PF; PrcePFF.. R**GAMMAF*RL**(-GAMMAF)=G= PFF; E_TMZ.. PFX*PMZ=G=PZ; E_TEZ.. PZ=G=PFX*PEZ; E_TMY.. PFX*PMY=G=PY; E_TEY.. PY=G=PFX*PEY; * Market clearng contons: W.. W*4000=G=CO/PW; Z *Z+299*TMZ=G=300*TEZ+ALPHA*PZ**(ALPHA-)*PY**(-ALPHA)*W*4000; Y.. 800*Y+300*TMY=G=299*TEY+(-ALPHA)*PY**(-ALPHA)*PZ**ALPHA*W*4000; X.. 200*X =G= (PX)**(-IGMA)*(ELTA*(PX)**(-IGMA)+(-ELTA)*(PXF)**(- IGMA))**(IGMA/(-IGMA))*P*400/2; XF.. 200*XF =G= (PXF)**(-IGMA)*(ELTA*(PX)**(-IGMA)+(-ELTA)*(PXF)**(- IGMA))**(IGMA/(-IGMA))*P*400/2; VA.. VA*400=G= B*PVA**(-THETA)*(PP**(-THETA)+PVA**(-THETA))**(THETA/(- THETA))*Y*800; P.. P*400=G= B*PP**(-THETA)*(PP**(-THETA)+PVA**(-THETA))**(THETA/(- THETA))*Y*800; RL.. L_EOW=G= (-GAMMAVA)*VA*400*R**GAMMAVA*RL**(-GAMMAVA)+(- BETA)*3200*Z*R**BETA*RL**(-BETA) +(-GAMMA)*X*25*(**(/(- IGMA))/C)*R**GAMMA*RL**(-GAMMA)+ (-GAMMAF)*XF*(F**(/(-IGMAF))/FC)*70*R**GAMMAF*RL**(-GAMMAF)+ (-GAMMA)**7.5*R**GAMMA*RL**(-GAMMA)+(-GAMMAF)*F*8*R**GAMMAF*RL**(- GAMMAF); R.. _EOW=G= GAMMAVA*VA*400*RL**(-GAMMAVA)*R**(GAMMAVA- )+BETA*Z*3200*RL**(-BETA)*R**(BETA-) +GAMMA*X*(**(/(- IGMA))/C)*25*RL**(-GAMMA)*R**(GAMMA-)+ GAMMAF*XF*(F**(/(-IGMAF))/FC)*70*RL**(-GAMMAF)*R**(GAMMAF-)+ GAMMA**7.5*RL**(-GAMMA)*R**(GAMMA-)+GAMMAF*F*8*RL**(- GAMMAF)*R**(GAMMAF-);.. FC**PF=G=ETRE; 32

41 Chapter. Appences F.. FCF*F*PFF+LT*F=G=ETREF; PFX.. TEY*PEY+TEZ*PEZ=G=TMZ*PMZ+TMY*PMY; *Income balance contons: ICO.. CO=E= L_EOW*RL+_EOW*R+LT*F+TFX*XF*F**(/(-IGMAF))*200/FC; IETRE.. ETRE=E=MK*(PX)*X*200; IETREF.. ETREF=E=MKF*(PXF)*XF*200; * hares harep.. HP =E= (P**((THETA-)/THETA))/((VA**((THETA-)/THETA))+(P**((THETA- )/THETA))); haresh.. sh =E= (ELTA*X**((IGMA-)/IGMA))/(ELTA*(X**((IGMA-)/IGMA))+(- ELTA)*(XF**((IGMA-)/IGMA))); hareshf.. sh+shf =E= ; *Markups Markup.. MK =E= ((/THETA)-(/IGMA)+HP*(-/THETA))*(v+(-v)*sh/)+ ((/IGMA)-(/IGMA))*(v+(-v)/)+(/IGMA)*((v*v+)/(+v*(-))); MarkupF.. MKF =E= ((/THETA)-(/IGMA)+HP*(-/THETA))*(vf+(-vf)*shf/F)+ ((/IGMA)-(/IGMAF))*(vf+(-vf)/F)+(/IGMAF)*((vf*vf+)/(+vf*(F-))); Moel Mo5 /PrceW.W, PrceZ.Z,PrceY.Y, PrceVA.VA, PrceP.P, PrceX.X, PrceXF.XF, PrcePF., PrcePFF.F, E_TEY.TEY, E_TMZ.TMZ, E_TEZ.TEZ, E_TMY.TMY, W.PW, Z.PZ, Y.PY, X.PX, XF.PXF, VA.PVA, P.PP, RL.RL, R.R,.PF, F.PFF, ICO.CO, IETRE.ETRE, IETREF.ETREF, harep.hp, haresh.sh, hareshf.shf, Markup.MK, MarkupF.MKF, PFX.PFX/; W.L=; Z.L=; Y.L=; PY.L=; VA.L=; P.L=;X.L=; XF.L=;.L=0; F.L=0; RL.L=; TEY.L = 0; TMZ.L = 0; TEZ.L = ; TMY.L =; PXF.L=; PZ.L=; PVA.L=; R.L=; PP.L=; PF.L=; PX.L=; PFF.L=; PW.L=; CO.L=4000; ETRE.L=50; ETREF.L=55; PZ.L=; MK.L=0.25; MKF.L=0.275; sh.l=0.5; shf.l=-sh.l; HP.L=0.5; C= /(.L**(/(IGMA-))*(-MK.L)) ; FC= (+TFX)/(F.L**(/(IGMAF-))*(-MKF.L)); pxf0=f.l**(/(igmaf-)); xf0=xf.l/(f.l*f.l**(/(igmaf-))); px0=*.l**(/(igma-)); x0=x.l/(.l*.l**(/(igma-))); *umerare: PFX.FX=; OLVE Mo5 UIG MCP; splay pxf0, xf0,px0, x0; *Counterfactual experment: o lumpsum value tax! LT=0; OLVE Mo5 UIG MCP; *Counterfactual experment: o output tax! LT=2.5; TFX=0; OLVE Mo5 UIG MCP; *Counterfactual experment: Free trae! TFX=0; LT=0; OLVE Mo5 UIG MCP. 33

42 Chapter 2 On Equlbrum tablty n the CGE moels wth Monopolstc Competton A smple general equlbrum moel wth monopolstc competton s constructe to stuy the contons of equlbrum stablty. The effects leang to unstable equlbrum an the subsequent counterntutve comparatve statcs result are entfe. Moreover, the mpacts of the parameters of the moel on the stablty of the benchmark equlbrum are presente by evelopng ynamc extensons unerlyng the ntally statc moel. In partcular, t s shown that the elastcty of substtuton between the xt-tgltz varetes, fference n the sklle labor ntenstes of two sectors of the economy an the relatve sze of the ncreasng returns to scale sector ncrease the stablty of the benchmark equlbrum. The ynamc tme paths of the enogenous varables from unstable to stable equlbrum are presente. It s also foun that the results of an a valorem tax cut woul have the same effect n terms of rectons of change on most of the enogenous varables of the moel rrespectve of ata. 2.. Introucton Ths chapter constructs a smple general equlbrum moel wth two factors of proucton an two sectors of the economy: the ncreasng returns to scale sector wth monopolstc competton market structure an the perfectly compettve constant returns to scale sector. It then stues the mpact of a tax cut n the ncreasng returns to scale sector on the output of the nustres an factor returns. It s foun that uner the small open economy assumpton, t s possble to classfy the results nto two categores n terms of the recton of change of the enogenous varables. However, small open economy assumpton also elmnates the prce effect whch makes the result epenent on nrect effects. Ths makes t possble for a counterntutve result to occur. We entfy nrect effects an stuy the parameters of the system that nfluence them. In recent lterature on nternatonal trae, the ncreasng returns to scale sector wth monopolstcally compettve market structure s assume to escrbe the servces sector (Markusen et al., 2005). Ths s plausble because there are many varetes of servces an a hgher number of them makes t more lkely that consumer specfc servces are avalable. Therefore, the avalablty of more varetes of servces ncreases consumer utlty (love of varety effect). It woul also be reasonable to assume that sklle labor s use more ntensvely by servces compare to other nustres. The conventon to gve some specal name to factors of proucton or sectors of the economy epens on the problem beng consere. In partcular, n the core-perphery moel of Krugman (99), the ncreasng returns to scale sector s assume to be manufacturng an the perfectly compettve sector s agrculture. Consequently, the manufacturng sector s assume to employ workers an the agrcultural sector s assume to employ peasants. Therefore, wthout a loss of generalty let us assgn sklle labor to be the factor that s ntensvely use n the monopolstcally compettve or servces nustry. It coul also be the manufacturng nustry whch s lkely to be subject to mperfect competton as well. 34

43 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Alreay n amuelson (947) t was state that the comparatve statc outcomes are ntmately te to the stablty of the equlbrum. Consequently, the stablty of the benchmark equlbrum nees to be stue n orer to etermne f the counterntutve result s stable or not. There are two stablty concepts epenng on the market ajustment mechansms: Marshallan an Walrasan stablty. In Walrasan equlbrum, market ajustment s efne n terms of enogenously changng prces an n Marshallan equlbrum t s efne n terms of quanttes. Each concept s chosen epenng on the nature of the economc settng. Even though the Walrasan stablty concept s often mple by efault n many stues whle checkng for equlbrum stablty, t s not sutable for our moel because we have fxe prces of fnal goos. Furthermore, the Marshallan stablty concept s assocate wth the theory of proucton an t s more approprate to the settng wth small open economy assumpton an corresponngly wth no ajustments n the prces of trae goos. There are several approaches to etermne the stablty of the equlbrum by stuyng alternatve specfcatons of the ynamc extensons of the statc moel. Gven that we have alreay establshe that the Marshallan stablty concept s approprate to our moel settng, the most straghtforwar way s to check t by lookng at whether the conton of Marshallan output ajustment process hols (Ie an Takayama, 990). Another way s to resort to the methos use n the new economc geography theory (Krugman, 99). In that case, the ynamcs of the labor movement between sectors s stue n analogy wth the nterregonal labor movement of new economc geography moels. In these moels, sklle labor s assume to be mmoble between regons n the short run. On the other han, unsklle labor s assume to be always mmoble. Ths enables us to uncover new nsghts of moel behavor an etect agglomeraton type effects. It also proves us wth the framework for ntroucng the ynamc evoluton of varables by stngushng between short run equlbrum where sklle labor s mmoble between the sectors an long run equlbrum where t s moble between the sectors. Fnally, the specfcaton wth the ynamcs n the number of varetes of the ncreasng returns to scale sector that also appears to trgger the counterntutve effect s stue. It enables us to obtan an analytcal representaton of the parameter mpacts on stablty. The results wth respect to the mpact of the structural parameters of the moel on stablty are foun to be entcal an the counterntutve result s unstable n all specfcatons. The nstablty of the equlbrum obtane from the benchmark ata makes us conclue that the counterntutve effect cannot be gven any reasonable nterpretaton. evertheless, a proper comparatve statcs coul be acheve subject to specfc assumptons even uner an unstable ntal equlbrum accorng to the global corresponence prncple of amuelson. In partcular, amuelson (97) emonstrate that the mpact of the nternatonal transfer s nepenent of the sze of the transfer an the stablty or nstablty of the ntal equlbrum. Bhagwat et al. (987) generalze the global corresponence prncple by extenng t to the shfts n any parameter of the system an not only nternatonal transfers. In other wors, n an economy wth two goos, the comparatve statcs of a shft n any parameter of the moel oes not epen on the stablty or nstablty of the ntal equlbrum or on the magntue of the shft f some certan contons hol. These contons are less restrctve than the requrement of Walrasan stablty. It shoul be note that Walrasan stablty ensures the applcablty of the stanar fferental calculus methos. That s to say, f equlbrum s unstable, the mpact of the parameter shft gven by the mplct functon theorem wll be 35

44 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton msleang. Consequently, one of the contons guarantees the exstence of a postve equlbrum prce an another s a ynamc prce ajustment mechansm or Walrasan tâtonnement. It postulates that the relatve prce of a goo (remember that we have only two goos) shoul ncrease (ecrease) when excess eman corresponng to that goo s postve (negatve). Hence, more general crtera than the Walrasan equlbrum stablty coul be obtane so that t s possble to have val comparatve statcs results even n the case when ntal equlbrum s unstable. Ie an Takayama (990) erve the same contons n the case wth ncreasng returns to scale an fxe worl prces. In that case, Marshallan stablty shoul be ensure. It s justfe to be equvalent to the stablty of the long run equlbrum. Furthermore, Ie an Takayama (990) state that many comparatve statc paraoxes assocate wth varable returns to scale sappear when Marshallan stablty s assume. mlarly to the case of Walrasan stablty, less restrctve assumptons requre for val comparatve statcs results coul be formulate. Consequently, the assumpton equvalent to the Walrasan tâtonnement woul be a Marshallan output ajustment process. Ths allows us to emonstrate the approprate comparatve statcs outcome of an a valorem tax cut from the unstable benchmark equlbrum pont of our moel. The rest of the chapter s organze as follows: secton 2 presents the settng of the base moel wth all of the equatons n the mxe complementarty format. It also escrbes ata an the calbraton proceure. ecton 3 presents the results of the moel an erves a proposton that classfes the results nto two categores epenng on changes n the nustres of the economy. It further nvestgates the stablty of the equlbrum an emonstrates proper comparatve statcs outcome. ecton 4 raws on analoges wth new economc geography lterature. ecton 5 erves the analytcal contons of stablty an full specalzaton by selectng the core equatons responsble for the moel behavor an ecton 6 conclues. In Appenx A, we reprouce both the counterntutve result an the mpact of the parameters uner a fferent ataset by transformng a stanar textbook CGE moel. Appenx B represents the ynamc tme paths of the enogenous varables an Appenx C contans the coe of the moel wrtten n the GAM syntax Moel structure 2.2. Base moel settng There are two goos, enote by Z an X, prouce n the small open economy. The consumer utlty nex s a Cobb ouglas composte of those goos an t s gven by: U Z X (2.2.) The X nustry, henceforth an IR (ncreasng returns to scale) nustry s a CE composte (constant elastcty of substtuton) of several varetes gven by: X n ( x j ) (2.2.2) j Here s the elastcty of substtuton between varetes:. Each varety s, n turn, a Cobb ouglas composte of sklle an unsklle labor: x L, n (2.2.3) 36

45 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Here, L are the quanttes of sklle an unsklle labor corresponngly, use n the proucton of varety. ubsequently, the ual prce nex wll be a composte of factor prces: c w v. Here w an v are the wages of sklle an unsklle labor. The nustry s subject to an a valorem tax t. Then, PX(-t) s an after tax prce face by an IR proucer. Each frm n the IR sector prouces one varety only an a monopolstc competton market structure s assume. Hence, the markup charge by frms wll be constant an gven by: markup The value of the nustry fxe cost s taken to equal the net markup revenue: FC * c* n PX * X *( t) (2.2.4) Here FC s a quantty of the fxe cost per frm an c s a margnal cost composte of factor prces; v The proucton of goo Z s subject to constant returns to scale, henceforth CR nustry, an perfect competton. In ths case, the proucton functon s gven as: Z L (2.2.5) I assume that the proucton of an IR nustry s more sklle labor ntensve than the proucton of a CR nustry:. The prces of both fnal goos PZ an PX woul practcally be fxe uner the assumpton of a small open economy an free trae n fnal goos. As t coul be seen from the expressons below, the prces of the goos coul only move together wth the value of the exchange rate: PX epx an PZ epz * * (2.2.6) * * Here e enotes an exchange rate an PX, PZ are the worl prces of both goos. v In other wors, export an mport prces of both goos wll be fxe mplyng fxe prces of the fnal goos. Uner the symmetry assumpton, (2.2.2) woul transfer nto: X x* n x* n* n (2.2.7) It means that the total nustry output wll be hgher than the output of all of the varetes: X x* n ; Ths s explane by the love of varety effect. Let me enote the corresponng multpler as: n n. Then, the nustry prce PX woul be fferent from the varety prce px: px PX (2.2.8) n The prce of an nvual varety wll epen on the IR nustry prce an love of varety multpler ( n ). Accorng to the small open economy assumpton, the nustry prce s fxe an the love of varety multpler epens postvely on the number of varetes an negatvely on the elastcty of substtuton. v Assumng the same factor prce composte for fxe an varable costs, makes our moel more tractable. v In the mxe complementarty form of moel equatons, n orer to have a possblty for both goos to be ether exporte or mporte, I stngush between export worl prce an mport worl prce. Therefore, ntally nactve equatons are ae to the coe. 37

46 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Equatons of the moel n the mxe complementarty problem format The equatons of the moel are wrtten n the mxe complementarty problem (MCP) format whch s use to express fferent types of economc moels. Thus, the moel equatons coul be ve as zero proft contons, market clearng contons an ncome balance contons n the stanar way. Uner the assumpton of symmetry between varetes, all of the equatons are gven at the nustry level. v Zero proft contons: In general, the followng conton shoul be satsfe for every sector: total cost total revenue, output 0, (total cost - total revenue) output = 0 The profts n the IR sector woul also be zero after we account for taxes an markup. Output s a complementarty varable assocate wth every zero proft conton. For X : c w v PX ( t) n( ) px( t)( ) (2.2.9) For Z w v PZ : (2.2.0) For W : PZ PX PW (2.2.) I assume that the goo Z s exporte an the goo X s mporte n the equlbrum. The exports of Z are enote TEZ an the mports of X are enote TMX: For TEZ : PZ epez (2.2.2) For TMX : epmx PX (2.2.3) The worl export prce of Z (PEZ) an the worl mport prce of X (PMX) are fxe. The equatons corresponng to the mports of goo Z an exports of goo X are also ae. They are calbrate to be nactve n the benchmark equlbrum by takng the prces PEX an PMZ slghtly fferent from PMX an PEZ: For TEX : PX epex (2.2.4) For TMZ : epmz PZ (2.2.5) Market clearng contons: The general form s: supply eman, prce 0, (supply - eman) prce = 0 The prce of the commoty s an assocate varable here. We shoul, therefore, conser each exstng commoty n turn. For PW : W * PW CO (2.2.6) v They coul also be gven at the varety level as we o n the other verson of the same moel. It s mportant to ensure that f the rght han se of the equaton s gven n quanttes or values at the nustry level, then the same shoul be true for the left han se of the equaton. All those equatons wth excepton of calbraton multplers are present n the program coe. 38

47 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Here PW s a utlty prce nex. It coul also be thought of as an expenture functon; CO s an ncome of the representatve consumer. PW For PZ : Z TMZ TEZ W (2.2.7) PZ PW Accorng to the hepar s lemma, represents the eman for goo Z by the PZ representatve consumer per unt of utlty. mlarly for X: PW For PX : X TMX TEX W (2.2.8) PX Let us enote the enowments of sklle an unsklle labor n the economy as an L. Then the market clearng contons for the factors of proucton coul be gven as follows: PZ c X c For v : L LZ LX Z FC n (2.2.9) v v n v PZ c X c For w : Z X Z FC n (2.2.20) w w n w The prmary factors are emane by both sectors an, L are the amounts of sklle an unsklle labor emane by sector. The FC s gven at the quantty level an the latter term of equatons (2.2.9) an (2.2.20) enotes the amounts of the factors of proucton neee for the fxe cost. I assume that the factors of proucton are requre n the same proportons to cover fxe cost as for the proucton of the nvual frm. It shoul be X remembere that: x* n n ; For e : TEX * PEX TEZ * PEZ TMX * PMX TMZ * PMZ (2.2.2) The equaton (2.2.2) s a balance of trae wth the exchange rate as an assocate varable. Income balance conton: A general form of the ncome balance conton s: ncome enowment tax revenue, ncome 0 ; In our moel, only the ncome of the representatve agent s consere: For CO : CO w Lv tpx * X (2.2.22) The last expresson equates the total fxe cost to the markup revenue an has n as an assocate varable: x For n : FC * c* n* PX *( t)* X (2.2.23) (2.2.4) ata an calbraton The mcro-consstency matrx looks as follows: x The zero profts n the X nustry are attane by changes n the number of frms. Therefore, to be more precse, the last equaton coul also be thought of as the actual zero proft conton for X. The former zero proft conton woul then be classfe as a proft maxmzaton conton equatng margnal cost to margnal revenue. 39

48 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Table 2.. Mcro-consstency matrx Representatve Proucton sectors Consumer Markets Z X TEZ TMX U CO PZ PX PW v w e A valorem tax (t) Fxe cost The number of frms s taken to be equal 0 an the a valorem tax (t) s The benchmark values of prces an quanttes are calbrate to be one for easer comparson wth the results of the counterfactual experments. The elastcty of substtuton s calbrate accorng to the value of the fxe cost. In partcular, for the mcro-consstency matrx above: 5. The shares of the factors of proucton are taken as follows: 0.725, 0.75, 0.3 ; If not specfe fferently, exchange rate e s taken to be the numerare Results from the base moel an equlbrum stablty 2.3. Interpretaton of the results Uner the ownwar slopng eman for an IR goo, a fall n the a valorem tax woul result n a lower IR nustry prce an the expanson of the proucton. There s no such prce effect generate by the ownwar slopng eman n our moel ue to the perfectly elastc eman for fnal goos. However, t coul be moele, for example, by assumng a close economy or mperfect substtutablty between mports an omestc goos an an mperfect transformablty between exports an omestc goos. Otherwse, the nrect effects wll etermne the changes n the IR sector output. Consequently, accorng to (2.9), a fall n the a valorem tax woul lea to ajustments n () the number of frms of the love of varety multpler (economes of scale effect), or n (2) the prces of the factors of proucton (crowng n the market for sklle labor effect). Those two ajustments woul be pushng the proucton of the IR nustry nto two fferent rectons. Consequently, the values of the parameters nfluencng the relatve strength of the two ajustments woul etermne whether the IR nustry proucton expans or contracts. It s possble to show that t s enough to know the recton of change (rse or fall) n the output of one of the two sectors to etermne the rectons of change of the other enogenous varables after a shft n some exogenous varable. Wthout loss of generalty we coul assume an IR sector expanson to take place an analyze the changes n the enogenous varables of the moel after some change n the a valorem tax. Let me use the hat notaton to enote the percentage change of a varable as follows: ˆ X X. Then, X Xˆ 0 ( Xˆ 0) f X s ncreasng (ecreasng). 40

49 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton It coul be shown that after any change n an IR nustry output, the frm (varety) level output wll be unchange: xˆ 0. If we plug c from (2.2.9) nto (2.2.4), we coul obtan the followng expresson: PX ( t) n( ) nfc PX ( t) X (2.2.4) or X FCnn ( ) (2.4) x FC( ) xˆ 0. Therefore, hgher elastcty of substtuton woul make t more lkely that every frm woul prouce more output because there s less nee to prouce more varetes of goos for the gven level of nustry output when varetes are more substtutable between each other. mlarly, a hgher fxe cost makes t more optmal to prouce fewer varetes to cut the total value of fxe cost an corresponngly more of each varety for a gven value of nustry output. Apparently from (2.2.7) t follows that: X ˆ ˆ ( ˆ 0 ˆ 0) n X n. mlarly, px 0 follows from (2.2.8) an PX PZ 0 shoul hol uner a small open economy assumpton. Let us make some further transformatons to the moel equatons n orer to have an easer nterpretaton of the effects. The market clearng contons for both factors of proucton can be transforme as follows: PZ c X c ( w * v ) ( w * v ) X Z FCn Z FC * n w w n w w w n PZ c X Z FCn (2.3.) w w n ow, f I plug c from equaton (2.2.9) nto (2.3.) an multply t by w, we coul erve the followng: X X ( ) ( ) w PZ * Z PX ( t) n( ) FCn PZ * Z PX ( t) FCnn n X ( ) X PZ * Z PX ( t) PZ * Z PX ( t) X w PZ * Z PX ( t) X (2.3.2) mlarly for v: vl ( ) PZ * Z ( ) PX ( t) X (2.3.3) In the hat notaton, t wll look as follows: wˆ shx Xˆ ( shx) Zˆ (2.3.4), vˆ shx2 Xˆ ( shx2) Zˆ (2.3.5) PX ( t) X ( ) PX ( t) X where shx, shx2. PZ * Z PX ( t) X ( ) PZ * Z ( ) PX ( t) X Moreover, a zero proft conton for CR sector (2.2.0) coul be wrtten as: wˆ ( ) vˆ 0 or wˆ ( ) vˆ (2.3.6). In other wors, sklle an unsklle wages woul be changng n fferent rectons. As a result, when the sklle labor ntensve IR sector expans the unsklle labor ntensve CR sector contracts: Xˆ 0 Zˆ 0.It coul be prove by contracton: f Z ˆ 0, then from (2.3.4) an (2.3.5) we obtan wˆ 0 an vˆ 0 whch contracts (2.3.6). 4

50 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Inee, t cannot be another way aroun wth no factor movements between the small open economy an the rest of the worl an wth fxe enowments of both factors of proucton, perfectly nelastcally supple to corresponng factor markets. ubsequently, wth expanson n the IR sector, the economy specalzes more n the proucton of an IR goo an starts mportng/reuces ts exports of a CR goo: XX 0, XZ 0, XX TEX TMX, XZ TEZ TMZ Even though the expanson of the IR sector woul brng a hgher eman from the IR sector for both factors of proucton, the total eman n the economy for sklle labor wll ncrease an the total eman for unsklle labor wll fall. In other wors: wˆ 0 an vˆ 0. Ths coul also be prove by contracton. Let us assume that wˆ 0, then vˆ 0. For these nequaltes to hol an gven that X ˆ 0, Z ˆ 0, shx shoul be lower than shx2: shx shx2 (2.3.7) ( shx2) ( shx) (2.3.8) ( ) PZ * Z ( ) PX ( t) X Let me enote 0. PZ * Z PX ( t) X Then equatons (2.3.7) an (2.3.8) coul be rewrtten as: (2.3.9) (2.3.0) From (2.3.9) an (2.3.0) t follows that:. Ths contracts our assumpton of wˆ 0. Therefore, more eman for sklle labor an less eman for unsklle labor woul brng a hgher prce of the former an a lower prce of the latter. The total mpact on the margnal cost composte c shoul therefore be ambguous. However, t s easy to prove that the composte wll ncrease: Usng the prevous results of wˆ 0 an vˆ 0 an assumpton that whch s equvalent to, we coul wrte own an equaton (2.3.6) n the followng way: wˆ wˆ ( ) vˆ ( ) vˆ wˆ ( ) vˆ 0 cˆ 0. Usng the same lne of reasonng, entcal changes n the varables but only wth a reverse sgn coul be obtane for X ˆ 0. Consequently, we have prove the followng proposton: Proposton : Let us assume that contons (2.2.9) an (2.2.0) hol wth equalty (.e. X 0, Z 0 ) an. Then, any change n the a valorem tax t that leas to (a) an ncrease n the IR sector output ( ˆ X X 0 ) (b) a ecrease n the IR sector output X ( X ˆ 0 ) woul also brng the followng changes n the enogenous varables of the moel: ( a) xˆ 0, nˆ 0, px 0, Zˆ 0, wˆ 0, vˆ 0, XX 0, XZ 0, cˆ 0 ; ( b) xˆ 0, nˆ 0, px 0, Zˆ 0, wˆ 0, vˆ 0, XX 0, XZ 0, cˆ 0 ; 42

51 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton mlar versons of Proposton coul be formulate for other exogenous varables. In partcular, f we ntrouce a per unt of output tax nstea of an a valorem tax, we coul erve the same changes n the varables except xˆ 0,( 0) wth X ˆ 0,( 0). Table 2.2 presents the results of the moel n the form of the percentage change from the benchmark n enogenous varables after a reucton n the a valorem tax. The results for the ata escrbe n secton 2.3 are gven n the ntal case column. Table 2.2. Percentage change from the benchmark after a tax cut (t = 0) Varables: Intal case 0 0 ( ) 0.65 ( ) Welfare (W) -.7% 5.8% 3.% Perfectly compettve sector (Z) 5.5% -3.% -9.8% IR nustry (X) -59.7% 300.9% 05.8% umber of frms (n) -5.7% 248.9% 78.% et exports of Z (TEZ-TMZ) 75.6% % -33.6% et mports of X (TMX-TEX) 75.6% % -33.6% Prce nex for welfare (PW) 0.0% 0.0% 0.0% Prce of the Z goo (PZ) 0.0% 0.% 0.% Prce of the IR nustry (PX) 0.0% -0.% x -0.% Payments to the other factors 3.3% -6.4% -6.% of proucton (v) Payments to the sklle labor (w) -7.3% 52.4% 43.6% Exchange rate (e) 0.0% 0.0% 0.0% Consumer ncome (CO) -.7% 5.8% 3.% Prce of the nvual frm (px) -6.6% 4.7% 5.4% Output per frm (x) 0.0% 0.0% 0.0% Inlne wth the statements n Proposton (b), an IR sector contracton brngs a fall the number of frms an the country wll mport even more of the IR goo X an prouce more an export more of goo Z. If we take fferent values of the parameters as ncate n the secon an thr column of Table 2.2, the outcomes woul be nlne wth Proposton (a). Actually, Proposton explans almost all of the rectons of change n the enogenous varables. On the other han, the magntues of the changes wll epen on the values of the parameters an ata n the mcro consstency matrx. Moreover, because of the asymmetry n the two sectors an the economes (seconomes) of scale assocate wth the expanson (contracton) of the IR sector, the consumer welfare wll ncrease (ecrease) even for 0.5 ( 0.5). It s mportant to etermne the factors that lea to such fferent outcomes as n column one versus columns two an three or equvalently as n (a) an (b) of Proposton. After a fall n the a valorem tax, equaton (2.2.9) coul ajust ether through the eclne n the number of varetes (seconomes of scale ajustment) or ncrease n the factor composte c (market crowng ajustment). o, the frst ajustment s assocate wth the fall n the IR sector or (b) of Proposton an the secon wth the ncrease n the IR sector or (a) of Proposton. We have foun that c s ncreasng (ecreasng) wth the expanson (contracton) of the IR x The prces of the fnal goos are fxe. The mnor changes appearng n the last two columns of table 2 occur only because the prces of ntally nactve equatons PEX an PMZ are specfe to be slghtly fferent from PMX an PEZ. Ths s neee to pn own the net exports of both goos to some certan values. 0 43

52 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton sector. Consequently, both effects wll be opposte to each other n ether scenaro. Before testng the stablty of the obtane counterntutve result, let us stuy the mechansms that lea to ts appearance. The parameters that nfluence market crowng ajustment are sklle labor ntenstes of proucton or the cost shares of sklle labor n both nustres. In partcular, the hgher cost share of sklle labor n the IR sector () wll strengthen the ajustment. On the other han, snce the expanson of one nustry leas to the contracton of the other ue to the fxe values of factor enowments, hgher cost share of sklle labor n the CR sector ( ) wll weaken ths ajustment. In other wors, hgher value of brngs t closer to gven that. Ths means that the mpact of any change n the proucton mx of goos X an Z s less sgnfcant on the prces of the factors of proucton. Ths mples weaker market crowng ajustment. However, the results of the benchmark calbraton clearly emonstrate the omnance of ajustments n the number of frms (). A fall n the number of frms woul brng a fall n the proucton of the IR sector an the so calle contracton effect observe for benchmark ata. In ths case, a fall n the a valorem tax woul lea to the reshuffle n the IR nustry. We coul thnk of t as a stuaton n whch some frms woul ext the market because lower prces woul not brng a postve prce effect that normally results from hgher eman. In other wors, uner lower per output proucton costs an fxe prce of fnal goo, t woul be more effcent for the IR nustry to prouce fewer nvual varetes. Ths woul result n the contracton of the IR nustry gven the assumpton of a fxe nustry level prce. Ths ajustment s more lkely to occur wth hgher values of the love of varety multpler an therefore wth lower elastcty of substtuton between varetes ( ). Returnng to the results n Table 2.2, we observe that the overall ecrease n the factor composte s smaller than the fall n the love of varety multpler n the ntal case. The stuaton coul be reverse by ether ncreasng the elastcty of substtuton whch woul mnsh the love of varety multpler or ncreasng the fference n the sklle labor ntenstes of two sectors ( ) whch woul strengthen the factor market crowng effect. Ths s exactly what we o n the two last columns of Table 2.2. Both changes lea to the IR sector expanson an to a reverse n the pattern of trae n fnal goos. Ths s nlne wth the stanar prce effect an market crowng effect omnates the seconomes of scale effect. The outcomes of the ajustment n equaton (2.2.9) workng through the change n the prce of sklle labor wll be nlne wth the results n Markusen et al. (2005). In partcular, the servce trae lberalzaton brngs hgher wages for sklle workers ue to the expanson n the servces nustry. espte the rse n the number of frms, the rse n the factor prces composte s greater than the rse n the love of varety multpler. All nrect effects shoul have much less sgnfcance n the presence of prce effect. However, they coul stll explan some changes n the varables. In orer to stuy the mpact of a tax cut wth the prce effect, we coul, for example, conser a close verson of the moel. A close verson of our moel woul only mply the absence of the external sector. All the other parameters an functonal forms are the same as n the ntal case. The results of a close moel wth fferent numerare values are presente n Table

53 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Table 2.3. Percentage change from the benchmark after a tax cut for the close verson of the moel (t = 0) Varables/umerare PZ PX PW v w Welfare (W) 2.9% 2.9% 2.9% 2.9% 2.9% Perfectly compettve sector (Z) -.3% -.3% -.3% -.3% -.3% ownstream nustry (X) 4.7% 4.7% 4.7% 4.7% 4.7% umber of foregn frms (n).6%.6%.6%.6%.6% Prce nex for welfare (PW) -4.0%.5% 0.0% -3.3% -5.7% Prce of the Z goo (PZ) 0.0% 6.2% 4.2% 0.7% -.7% Prce of the ownstream nustry -3.9% 0.0% -0.3% -3.3% -5.4% (PX) Payments to the other factors of -0.7% 5.3% 3.4% 0.0% -2.5% proucton (v) Payments to the sklle labor (w).8% 8.2% 6.0% 2.5% 0.0% Consumer ncome (CO) -.3% 4.7% 2.9% -0.5% -3.0% Prce of the nvual frm (px) -.5% 2.8% -7.8% -0.9% -3.0% As a result of the prce effect, the mpact of a tax cut on the real varables of the moel s the same uner any numerare value. Apparently, the rect prce effect omnates over all the other effects an sets the recton of change n all the real varables. Consequently, an IR sector expans, a CR sector contracts an the representatve consumer attans a hgher utlty level ue to the economes of scale n the IR proucton. Even though the moel became smaller, t now comprses changes n the prces of goos. Therefore, the prces of the factors of proucton woul not always be changng n the fferent rectons as we ha n Proposton uner a small open economy assumpton. On the other han, changes n prces are relatve an epen on the numerare. In partcular, even though the prce charge by an nvual frm n the IR sector s ecreasng after a tax cut, t wll be slghtly ncreasng relatve to the prce of goo X. The case wth the prce of goo X as a numerare s specal because t s changng more n ts magntue than any other prce varable. Hence, the changes n prces coul swtch to be postve or sgnfcantly ncrease when expresse n terms of the PX. Ths s only a matter of the numerare choce an has no mpact on the moel behavor Equlbrum stablty an unqueness The counterntutve result obtane uner seconomes of scale mght be unstable accorng to the global corresponence prncple of amuelson. As we have alreay scusse n the ntroucton an accorng to Ie an Takayama (990), the Marshallan output ajustment mechansm shoul hol for any comparatve statcs outcome. Consequently, let us aopt the PX ( t) followng notaton: P s a relatve after tax worl prce of goo X an P s a PZ relatve supply prce or prce at whch proucers break even. It coul be erve from equatons (2.2.9) an (2.2.0): w P (2.3.) v n ( ) The relatve output of the IR sector wll epen on the supply prce. We coul nvert ths X relatonshp an assume that the supply prce epens on the relatve output: P Z. 45

54 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton The corresponng supply prce curve wll not be a supply curve as t only epcts zero proft break even ponts an coul be both upwar an ownwar slopng. Then, the followng ynamcs coul be mple: 0 f P P X P X a 0 f P P (2.3.2) Z P Z 0 f P P Here a s a postve constant. X X Let me enote q t, P( t) P ; Then, for ntal benchmark values we have: Z Z X 0 X q t, P( t ) P 0. Z Z The Marshallan output ajustment conton coul therefore be expresse n the form of the followng nequalty: X X X X X X q t, * P( t ) P * 0 (2.3.3) Z Z Z Z Z Z Here t X s the value of an exogenous parameter after a counterfactual experment an Z s a corresponng value of the relatve output of goo X. Furthermore, snce t s an a valorem tax, we assume that: t 0. In our moel, the frst factor of (2.3.2) s postve an the secon s negatve. Moreover, as we have seen from Proposton, a eclne n X s always renforce by ncrease n Z. Thus, the benchmark equlbrum s unstable. Therefore, takng the nstablty of the counterntutve result nto account, we coul formulate a stronger verson of Proposton : Proposton : Let us assume that the moel of (2.2.2) has an nteror soluton an. Then, an a valorem tax cut/rse woul brng the changes of the enogenous varables outlne n pont (a)/pont (b) of Proposton rrespectve of the moel ata an parameter values. In orer to obtan the correct response of the relatve output of the IR sector after a reucton n the a valorem tax, we coul assgn a hgher benchmark value of the relatve output wthout any change n the calbraton proceure uner the counterfactual experment. x X. L Inee, for 55 we eventually attan the proper response n the relatve output as Z. L gven n Table 2.4. x The program searches for an equlbrum soluton n the vcnty of the benchmark values of the enogenous varables. Those values are assgne before each moel calculaton. Consequently, t makes sense to search for a stable equlbrum at the pont wth a hgher than benchmark output of an IR goo. 46

55 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Table 2.4. Percentage change from the benchmark after a tax cut an fferent benchmark value of the IR sector output Welfare (W) 26.5% Payments to the other factors of proucton (v) -3.4% Perfectly compettve -68.3% Payments to the sklle labor 42.% sector (Z) (w) IR nustry (X) 786.2% Consumer ncome (CO) 26.5% umber of frms (n) 472.8% Prce of the nvual frm (px) 54.6% et exports of Z (TEZ- TMZ) -987.% et mports of X (TMX-TEX) % In orer to have a broa pcture of the processes n the moel, we coul plot the values of the relatve output of sector X for fferent values of the a valorem tax uner two fferent sets of benchmark values. The frst set of values s the same as our ntal benchmark values to whch the moel s calbrate an t assgns manly the values of one to prces an quanttes. Our alternatve set of values ffers only n assgnng a hgher than benchmark value to the output of goo X uner the same calbraton. Uner ths set of values we coul have a comparatve statcs outcome that satsfes the Marshallan output ajustment process. The curves obtane n ths way coul then be combne n one graph as emonstrate n Fgure 2.. Fgure 2. Moel stablty,25,000 E E 3 P P 0,875 E 0 E 2 P 0 0, X/Z X Hence, we coul thnk of t as a graph of P. It coul only be rawn numercally Z because there s no analytcal representaton for t. The relatve after tax prce of goos X splaye on the vertcal axs, changes only wth the value of the tax an P 0 an P are gven as follows: PX *( ) P t PX 0, P. The benchmark equlbrum s gven by the pont E 0. PZ PZ 47

56 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Consequently, the counterntutve effect consttutes the movement from the benchmark pont to pont E. However, t s msleang because E0 s unstable an an IR proucer woul have an ncentve to ncrease the proucton of goo X. Any small ncrease leas to postve profts for the IR proucer. Therefore, after any small perturbaton n the output, the economy woul move to the stable equlbrum pont E 2. In contrast to pont E 0, any small evaton from pont E2 woul brng us back to t. onetheless, as we have argue n the ntroucton, even f we start from pont E 0, t s stll possble to acheve proper comparatve statcs results. In partcular, movement from pont E 0 to E 3 satsfes the Marshallan output ajustment conton an nequalty (2.3.3). The changes n the other enogenous varables corresponng to the movement are presente n Table 2.4. Apparently movement from the pont E0 all the way to E3 brngs sgnfcant changes n the varables as coul be seen from Table 2.4. Thus, the correct comparatve statcs outcome consttutes of two effects: () change from unstable to stable equlbrum E0 E2 an (2) effect of the tax cut E2 E3. The comparatve statcs of change n the a valorem tax coul be calculate by applyng the theorem of mplct fferentaton on the equatons of our moel. However, t woul also result n the counterntutve movement E0 E because t s applcable only subject to the stablty of the ntal equlbrum. Consequently, the Marshallan output ajustment process represents more general contons that secure val comparatve statcs outcomes. The curve n Fgure 2. coul ben further ownwars as t passes pont E3 because of another set of approprate benchmark values. Ths woul nclue a fferent range of values that coul brng another counterntutve response n the frm s output. Consequently, n aton to the problem of stablty, there coul also be a problem of the non unqueness of the equlbrum. There s no unversal test to verfy the unqueness of the equlbrum an apple general equlbrum stues often presume that the result obtane from the moel s unque. Wng (2004) surveys several specal cases where unqueness can be assure uner the assumptons on the stortons mpose by counterfactual experments an functonal forms. In partcular, Mas Colell (99) proves that for the so calle super Coub-ouglass economes the equlbrum wll be unque n the absence of stortons or f those storton such as taxes are exogenous an o not epen on the enogenous varables of the moel. The super Coub-ouglas economes are the moels wth super Coub-ouglas proucton an utlty functons an the CE functons wth the elastcty of substtuton greater than or equal to one are a partcular case of such functons. Consequently, n our moel wth exogenous tax stortons an super Coub ouglas economy we coul assert that the stable equlbrum E 3 s unque Alternatve perspectve on stablty In ths secton the stablty s stue from the perspectve of the ew Economc Geography (EG) moels. Ths approach s sutable because the agglomeraton effects obtane n EG moels often stem from IR sectors. mlarly, the expanson or contracton of a small open economy after trae lberalzaton an accompanyng specalzaton n one of the sectors n our moel s analogous to the agglomeraton effects n EG moels. Consequently, we coul expect smlar ynamcs to occur n our moel an we coul use the same tools to stuy stablty. Moreover, ths specfcaton enables us to obtan ynamc tme paths of the movement of enogenous varables from unstable to stable equlbrum by ntroucng ynamc conseratons. In ths specfcaton, the stablty analyss wll be base on the 48

57 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton ncentves of hgh sklle workers to move to a sector (regon) wth a hgher relatve wage. Therefore, n orer to erve the result nterpretaton whch s equvalent to the EG moel results, we wll nee to ntrouce some transformatons to our moel. In the semnal paper of Krugman (99), the xt-tgltz base ew Economc Geography moel has been evelope. The moel features 2 regons an 2 sectors, such that, smlarly to our moel, one sector (agrculture) s governe by constant returns to scale an another sector (manufacturng) s subject to ncreasng returns to scale proucton an s a xt- tgltz composte of varetes. However, the factors of proucton are assume to be sector specfc whch makes t fferent n terms of effects leang to agglomeraton an also analytcally more tractable than our moel. Currently ths moel s also known as a Core- Perphery moel (Fujta et al., 999) where a two regon economy coul have equlbrum wth a manufacturng core an agrcultural perphery or one where the proucton of two sectors s equally ve between regons. The core-perphery concentraton outcome occurs ue to the presence of backwar an forwar lnkages whch unerpn centrpetal forces. In partcular, a hgher varety of goos prouce n one regon lowers the prce nex an strengthens the ncentve to move more manufacturng nto that regon (forwar lnkage). On the other han, hgher ncomes n one regon woul lea to hgher wages brngng more workers there (backwar lnkage). There wll be no such effects n our moel because the supply of labor n the gven country s fxe an not moble wth the rest of the worl. Beses, the IR nustry prce nex s fxe an that preclues the prce effect. Whle real wages matter n the Krugman moel, n our moel t s suffcent to stuy nomnal wages. The ownwar pressure on the IR prce nex from an ncrease n the number of varetes s neutralze by an ncrease n the prce charge for each varety. Therefore, the agglomeraton effects of our moel occur as a result of the concentraton n one sector of the economy an the corresponng movement of both types of labor nto that sector. Consequently, let us assume two fferent sector specfc types of sklle labor. Ths assumpton s ntutvely reasonable an enables us to analyze stablty from the perspectve of the EG moels. The followng changes wll be ntrouce to the moel equatons: Frstly, sector specfc sklle labor woul now have fferent wages: c w2 v (2.4.) Therefore zero proft contons for both sectors woul be as follows: For X : w2 v PX ( t) n ( ) (2.4.2) For Z : w v PZ (2.4.3) econly, there wll be two market clearng contons for sklle labor corresponng to the two sectors of the economy: PZ For w: ( ) Z (2.4.4) w c X c For w2 : FCn (2.4.5) w2 n w2 Here s a share of the total enowment of sklle labor whch s specfc to the IR sector. 0 The benchmark value of ( ) s calculate from the share of sklle labor n the 49

58 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton IR sector of the ntal ataset. All the other equatons of the moel are the same as n the ntal case. The stablty s etermne by stuyng the ncentves of sklle workers to move to one of the sectors, represente by the wage rato, when the share of sklle workers n the IR sector ( ) changes exogenously. Fgure 2.2 features the wage rato for fferent values of uner the ntal scenaro. Fgure 2.2 Moel stablty from EG perspectve: Intal scenaro E0 E w2/w 0 0,05 0,2 0,35 0,5 0,65 0,8 0,95 q The ntal equlbrum pont E0 correspons to the benchmark value of an the benchmark 0 w2 wage rato of one:. It s unstable because any small ncrease n the ntal share of w sklle labor enowment of the IR sector woul result n the movement of sklle labor to that sector an away from pont E 0 n the long run ue to the hgher relatve wage. On the other han, equlbrum pont E s stable because any small evatons n woul generate wage ncentves for sklle labor to return back to pont E. In partcular, any ncrease (ecrease) n woul result n a lower (hgher) relatve wage of the IR sector an consequently more ncentves for sklle workers to move out of (move nto) the IR sector, that s back to E. Hence, the followng a hoc ynamc formulaton coul be assume: w2 *( ) (2.4.6) w Here 0 s some postve parameter. The competton between the sklle workers effect mentone n the Krugman moel or the market crowng effect mples that workers n the sector wth less labor wll face less competton an therefore hgher wages. In fact, ths effect wll be stronger wth the hgher n our moel. 50

59 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Fgure 2.3 Moel stablty from EG perspectve: s = 45 > s 0 E0 w2/w 0 0,05 0,2 0,35 0,5 0,65 0,8 0,95 q The mpact of on stablty s smlar to the one escrbe n the Krugman moel an our ntal settng. In partcular, Fgure 2.3 llustrates the equlbrum pont of the moel calbrate for the hgher than benchmark value of. In ths case, the wage rato curve s ownwar slopng an therefore E 0 s stable. Hgher elastcty of substtuton ecreases the economes of scale an consequently the ncentve to move nto the IR sector wth hgher values of. The market crowng effect between sklle workers wll be a omnant effect n that case. Fgure 2.4. Moel stablty from EG perspectve: g-β = 0.65 > g 0 -β 0 E0 E w2/w 0 0,05 0,2 0,35 0,5 0,65 0,8 0,95 q Another parameter etermnng stablty s the fference of sklle labor ntenstes of two sectors. From Fgure 2.4 we coul see that E 0 remans unstable but stable equlbrum E moves closer to E 0 uner a hgher than benchmark value of.on the other han, a lower than benchmark values of woul result n the alternatve stable full specalzaton equlbrum E (Fgure 2.5). Actually, lower values of such that woul make t easy to prouce any mx of two goos snce they woul requre the same mx of factors of proucton. Thus, gven the economes of scale n the IR sector, the country specalzes n proucng the IR goo only. Alternatvely, for hgh values of an uner the contons of full employment, t woul be more ffcult to change the ntal prouct mx. As a result, the scope of possble equlbrum values of shrnks as t coul be seen n Fgure

60 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Fgure 2.5. Moel stablty from EG perspectve: g-β = 0.< g 0 -β E w2/w E0 0 0,05 0,35 0,65 0,95 q In the Krugman moel, hgh ceberg transportaton costs make the concentraton equlbrum less lkely to occur. In orer to verfy ths effect n our moel, we coul ntrouce transportaton costs by assumng hgher mport prces of both goos: PMX For TMX : e PX (2.4.7) PMZ For TMZ : e PZ (2.4.8) Here 0 s an ceberg transportaton cost. Fgure 2.6 llustrates a sklle wage rato for 0.5. Intal equlbrum becomes stable an the effect s smlar to the Krugman moel. w2 It shoul be note that the benchmark wage rato changes to.934:.934 w Apparently after tarffs on both goos are remove, the economy woul ten to specalze n the proucton of one of the goos an trae the other one wth the rest of the worl. However, the resultng equlbrum pont coul be unstable. Therefore such a stuaton s less lkely to happen uner hgh transportaton costs. 0 Fgure 2.6. Moel stablty from EG perspectve: The case wth mport tarffs on both goos.,934 E0 w2/w 0 0,05 0,2 0,35 0,5 0,65 0,8 0,95 q 52

61 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Another parameter of the Krugman moel that cannot be elmnate by the choce of unts s the share of ncome spent on manufacturng goos. x There s no parameter corresponng to t n our moel. Even though we raw on the analogy between the labor movement across regons n the Krugman moel an the labor movement across sectors n our moel to stuy moel ynamcs, t s wrong to assume that the sectors n our moel are equvalent to the regons n the Krugman moel. evertheless, the Krugman moel wth full specalzaton woul be equvalent to the close verson of our moel. Only n that case coul we thnk about some equvalence between sectors an regons n the two moels. Otherwse the sectors are always asymmetrc. The monopolstc competton moel of nternatonal trae s also fferent because t leas to an ncrease n the number of varetes after lberalzaton of trae n goos whch s absent n our moel. We coul thnk of sklle labor to be moble between sectors n the long run. Then, s varable n the long run an every equlbrum for a partcular value of coul be thought of as a short run or nstantaneous equlbrum. In ths way we coul obtan ynamc tme paths of all of the enogenous varables. In partcular, from the screte tme verson of (2.4.6), coul be assume to ajust n tme accorng to the followng ynamcs: w2( T) ( T ) ( T) (2.4.9) w( T) Consequently, the ynamc evoluton of s llustrate n Fgure 2.7. X Z from an ntally unstable to stable equlbrum Fgure 2.7. ynamc tme path of the relatve output of goo X T X/Z x It s also taken to be equal to the share of manufacturng workers n the total number of workers employe n manufacturng an peasants employe n agrculture an enote by. That parameter wll be equvalent to the share of sklle workers: L n the lmtng case wth, 0 n our moel. Even n that case we cannot say that our small open economy an the rest of the worl are equvalent n any sense to the regon an regon 2 n the Krugman moel because sklle labor s mmoble between a small open economy an the rest of the worl. 53

62 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton We see that the output rato starts ncreasng at a rather slow rate whch rses n the meum term an subsequently subses as t reaches a stable equlbrum pont. In other wors, the output rato s growng at an ncreasng rate untl t reaches an nflecton pont somewhere n the meum term after whch t starts growng at a ecreasng rate. Other enogenous varables of the moel exhbt smlar ynamcs as t coul be seen from the Fgures n w2 Appenx B. The rato of sklle wages n two sectors represents the ncentve of sklle w workers to move to an IR sector an t equals one at the equlbrum. In other wors, there s no ncentve to move when the rato equals one. Fgure 2.8. ynamc tme path of the relatve wage of sklle labor n X sector,5,2,09,06 w2/w,03 0, T As emonstrate n Fgure 2.8, the ncentve to move s postve but small n the begnnng, reaches ts peak n the meum term an eclnes back to ts ntal level at the en of the ynamc path. We summarze that the EG perspectve confrme the nstablty of the equlbrum uner the ntal scenaro an enable us to fn agglomeraton type effects whch a) coul be opposte to an stronger than the market crowng of sklle workers effect b) coul lea to more specalzaton, not necessarly full specalzaton, of the economy n the IR sector. An mportant nsght of the EG moels s that cumulatve process begns when parameters of the system cross some crtcal values whch coul brng sgnfcant changes n the enogenous varables. It was the case wth the values of parameters an n our moel. The aopte approach also emonstrates how ynamc analyss coul be a powerful tool to smplfy an sort the outcomes of the statc analyss Analytcal expresson of stablty contons Alternatve moel formulaton In orer to erve analytcal expressons of stablty t s mportant to fn a reuce verson of the moel that woul preserve all the effects an relatonshps between the varables. uch a system oes exst for our moel. For that we nee to observe that the equatons assocate wth the external sector moelng such as (2.2.2), (2.2.3), (2.2.4), (2.2.5), (2.2.2) allow for any volume of trae as long as the trae balance hols an effectvely makes the prces of fnal goos constant. Therefore, we coul rop those equatons an assume constant prces of 54

63 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton goos to equal ther ntal calbraton values. Furthermore, uner the assumpton of a small open economy, the representatve consumer woul not nfluence the eman for fnal goos as there s an nefnte eman/supply of both goos from the rest of the worl. That woul not be the case, for example, n the verson wth the close economy. Consequently, the mpact on the welfare of the representatve consumer coul always be calculate resually from the changes n the outputs of the sectors. In other wors, we coul separate consumpton an proucton blocks an rop equatons relate to consumer welfare an market clearng contons for both goos. Thus, our moel reuces own to equatons (2.2.9), (2.2.0), (2.2.7), (2.2.8) an (2.2.4). Let us assume that the nteror soluton exsts an wrte own all of the contons n equalty form: ( ) w v Pn w v w Z P * X (2.5.) vl ( ) Z ( ) P * X FCw v n P * X The system (2.5.) efnes 5 equatons for 5 enogenous varables: w, v, X, Z, an n. The prce of goo Z s a numerare an P s a relatve after tax prce of goo X. The frst two equatons are zero proft contons for sectors X an Z, the next two are the market clearng contons for sklle an unsklle labor an the ffth equaton etermnes the number of frms by equatng the value of the fxe cost to the markup revenue. s a postve calbraton parameter use for choosng unts n a convenent way. If we take nto account that X * x * n an x FC *( ) we wll have: * *( )* (2.5.2) X FC n Therefore, the system coul always be reuce to 4 equatons wth 4 varables x. It shoul also be note that only equatons of (2.5.) were use to erve the contons of Proposton n secton 3. The analytcal conton of stablty coul be erve uner a more general case wth the CE compostes of the prmary factors of proucton. We assume Cobb ouglas compostes of the prmary factors of proucton n the frst an secon chapters of the ssertaton. Ths tme, let us assume that the elastcty of substtuton between sklle an unsklle labor s x an z n the IR an CR sectors respectvely. Then, the frm level proucton functons coul be wrtten as follows: x z x x z z x z x x z z x ( ) L, z ( ) L (2.5.3) The corresponng ual prce nexes wll be: x If we plug equaton (2.5.2) nto the last equaton of (2.5.) we wll obtan the frst equaton of (2.5.). 55

64 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton x x z x z z c w ( ) v, c w ( ) v z (2.5.4) x Ths woul transform the factor market clearng contons of (2.2.9) an (2.2.20): z x cz c : ( ) ( ) x X For v L Z FC * n (2.5.5) v v n z x cz cx X For w : Z FC * n (2.5.6) w w n Consequently, the man equatons of the moel wll be transforme as follows: ( ) cx Pn cz z x cz cx Z X Z P * X (2.5.7) x w w z x cz c x L LZ LX ( ) Z ( ) P * X x v v FCcxn P * X The values of X an Z coul be erve from the thr an fourth equatons of (2.5.7): X z z x x ( ) w v L v L ( ) w, Z (2.5.8) Pc ( ( ) w ( ) v ) c ( ( ) v ( ) w ) x z x z x z x z x z x z Apparently from the frst equaton of (2.5.), the calbraton parameter coul be chosen n 0 0 such a way that w v. Moreover, from the secon equaton of (2.5.7) we coul nfer that w v w v. Then, Z an X from (2.5.8) coul be wrtten as follows: X ( ) ( ) L, Z L (2.5.9) P( ) ( ) nce 0, the numerators n (2.5.9) shoul also be postve: ( ) L 0, L ( ) 0 (2.5.0) Those contons represent the constrants on factor enowments to be n the versfcaton cone efne by the technologcal parameters of the moel. The contons of (2.5.0) coul be combne nto the followng constrant on factor enowments: (2.5.) L The quanttes of sklle an unsklle labor coul be compare when w 0 v 0. Hence, conton (2.5.) states that the rato of sklle labor to the sum of sklle an unsklle labor shoul not excee the sklle labor ntensty of the IR sector. Otherwse t wll be mpossble to have full employment of sklle workers an the enowment pont (, L ) 56

65 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton woul be out of the versfcaton cone. mlarly, the rato of sklle labor to the sum of sklle an unsklle labor shoul not be less than the sklle labor ntensty n the CR sector because there wll be no full employment of unsklle workers otherwse. In orer to erve the stablty conton, let us ntrouce the followng varables: -unt cost rato of the IR goo over CR goo, r- the rato of sklle to unsklle labor w x z wage: r, k - factor nput rato for two sectors: kx, kz an - the rato of labor v Lx L z use n the IR nustry to total labor:. L Then, coul be etermne from the frst an secon equatons of (2.5.7): Lx cx P n ( ) (2.5.2) cz We coul use the hat notaton ( ˆ ) an the corresponng rules to present (2.5.2) n the followng form: nˆ ˆ cˆ ˆ x cz (2.5.3) x z w w If we assume that an, whch also hols x x z z w ( ) v w ( ) v for w v, the equaton (2.5.3) coul be wrtten as follows: 0 0 nˆ ˆ ( ) rˆ (2.5.4) Usng the fourth equaton of (2.5.7) an nsertng the value of P*X from the ffth equaton of (2.5.7), we coul obtan the expresson for as follows: x x cx cx ( ) P * X ( ) FCc x x x xn Lx v v cx FC n ( ) (2.5.5) L L L v x L Ths coul be transforme nto hat notaton: ˆ nˆ ( cˆ vˆ ) (2.5.6) x x Furthermore, f we use the expresson for cx from (2.5.4) an apply the hat transformaton we wll obtan the followng expresson for ˆ : ˆ nˆ ( wˆ ( ) vˆ vˆ ) nˆ ( wˆ vˆ ) nˆ rˆ (2.5.7) x x x Let us wrte the expresson for the rato of sklle to unsklle labor enowment usng (2.5.7): 57

66 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton x z x Lx z Lz kx ( ) kz (2.5.8) L L L Lx L Lz The factor nput ratos coul also be calculate from (2.5.7): x cx P * X x x x w w x z kx r, k (2.5.9) x z r Lx cx ( ) v ( ) ( ) ( ) P* X x v ubsequently, the total fferental of (2.5.8) coul be wrtten as follows: k k k k k 0 (2.5.20) x x z z z The values of the ervatves of kx an kz coul be foun from (2.5.9): x x k ˆ, ˆ x r r kx xr kz kz zr (2.5.2) ( ) If we plug expressons from (2.5.2) nto (2.5.20), we wll obtan the followng expresson: ( k k ) ( k ( ) k ) rˆ 0 (2.5.22) x z x x z z Let us now solve (2.5.22) for ˆ : ˆ kx x kz z rˆ (2.5.23) kx kz Here. Consequently, r an are monotoncally relate. The ntal equlbrum shoul be stable f an ncrease n the relatve wage of sklle labor (r) brngs an expanson of the nustry that uses sklle labor ntensvely. The equlbrum s thus stable f an only f ˆ 0 ˆ. In orer to obtan an expresson for n terms of the parameters of our moel, we ˆr ˆr coul nsert the value of ˆ from (2.5.23) nto (2.5.7) an then the obtane value of ˆn nto (2.5.4): kx x kz z nˆ ˆ x r an therefore kx kz kx x kz z rˆ ˆ ( ) rˆ x (2.5.24) kx kz After nsertng the values of kx an k z from (2.5.9) an uner w transforme nto the followng stablty conton: ˆ x z ( ) 0 (2.5.25) rˆ v, (2.5.24) s 0 0 Consequently, conton (2.5.25) coul be use to prove the followng proposton: 58

67 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton Proposton 2: Let us assume that the moel formulate n secton 2.2 has nteror solutons. The ntal equlbrum of the moel s stable f an only f ( x z )( ) (2.5.26) 2 ( ) Hence, the ntal equlbrum wll be stable for relatvely hgh values of an an for relatvely small values of. The last varable s small when the share of labor n the IR sector s hgh. Therefore, unstable equlbrum s more lkely to occur when the IR nustry s small. The conton (2.5.26) hols for any CR proucton functons an, an s are allowe to be functons of r. The Cobb-ouglas compostes of the prmary factors that we assume n our moel formulaton presente n secton 2.2 s just a partcular case of conton (2.5.26) wth constant, an x z : ( )( ) (2.5.27) 2 ( ) Contons of full specalzaton n the IR sector In Proposton 2, the mpact of the parameters on the equlbrum stablty has been prove. We have also foun that when the ntal equlbrum s unstable, t ncates that the moel wll evate from t so that the output of the IR sector wll ncrease relatve to the output of the CR sector. As a result of that the moel reaches a new stable equlbrum pont. nce t s not always easy to fn that new equlbrum, we may suspect that the pont at whch the economy specalzes n the proucton of the IR sector s the new stable equlbrum. It s mportant to fn out whether the full specalzaton n the proucton of the IR goo s equlbrum because t enables our small economy to realze the maxmum economes of scale. That, n turn, coul brng the hghest utlty to the representatve consumer as we have seen n secton 3. If full specalzaton n the proucton of an IR goo s an equlbrum, the revenue obtane from sellng only the IR goo subject to the resource constrants shoul be enough to pay the factors of proucton at ther ntal prce. In other wors, proucers wll have an ncentve to prouce the IR goo only f t brngs hgher profts. In terms of equatons n 0 0 (2.5.), the total expenture on the factors of proucton uner ntal prces w v s L. The last equaton of (2.5.) gves the value of total revenue of the IR 0 goo: P * X FC * n *. Consequently, we fn that full specalzaton outcome s not equlbrum after nsertng the benchmark values an comparng the two numbers: 0 P * X FC * n * 8*0* L Concluson How o we procee n the apple general equlbrum moels f the ntal equlbrum s foun to be unstable? Then, the equlbrum shoul be recalbrate by changng the values of the structural parameters. There are always parameter estmates whch are taken only wthn some certan range of values such as elastctes an factor ntenstes. Moreover, n our moel, t s possble to entfy the relatonshp between the parameter values an stablty of the ntal equlbrum. Consequently, t s possble to mplement the proper recalbraton proceure. It conssts of ncreasng the value of or the share of the IR nustry ( ) from (2.5.26) an /or ncreasng the value of. 59

68 Chapter 2. On Equlbrum tablty n the CGE moels wth Monopolstc Competton The counterntutve unstable outcome s more lkely to occur when the seconomes of scale effect s stronger than the factor market crowng effect n the absence of the rect prce effect. It also appears that the lower value of s ncreasng the possblty of havng a full specalzaton n the IR sector as an equlbrum uner suffcent economes of scale. The moel presents a goo example of how ynamc analyss coul be use to etermne the valty of the results an uncover new effects n the statc moel. Moreover, we have seen that even wth an unstable equlbrum t s stll possble to obtan val comparatve statcs. However, t wll not be meanngful for polcy analyss ue to the napproprately large changes n the enogenous varables. In that case, the changes n the enogenous varables coul be substantal because they ncorporate both the movement from unstable to stable equlbrum an the mpact from the change n the exogenous varable. Bblography Bhagwat, J., Brecher, R., an Hatta, T., (987) The global corresponence prncple: A Generalzaton, Amercan Economc Revew 77, pp Ie, T., an Takayama, A., (990) Varable Returns to cale an the Global Corresponence Prncple n the Theory of Internatonal Trae, Economc Letters 33. pp Fujta, M., Krugman, P., Venables, A.J. (999) The patal Economy Cambrge, MA: MIT Press. Krugman, P., (99) Increasng returns an economc geography, Journal of Poltcal Economy, 99: pp Markusen, J., (2002) General Equlbrum Moelng usng GAM an MPGE, Bouler. J., Rutherfor, T., an Tarr,., (2005) Trae an rect nvestment n proucer servces an the omestc market for expertse, Canaan Journal of Economcs 38 (2005), pp Mas-Colell, A., (99) On the unqueness of equlbrum once agan, n W.A. Barnett et al (es.), Equlbrum Theory an Applcatons: Proceengs of the xth Internatonal ymposum n Economc Theory an Econometrcs, ew York: Cambrge Unversty Press, pp amuelson, P., (947) Founatons of Economcs Analyss, Cambrge: Harvar Unversty Press; enlarge e., (983). amuelson, P., (97) On the tral of conventonal belefs about the transfer problem n J. Bhagwat et al., es., Trae Balance of Payments an Growth: Papers n Internatonal Economcs n Honor of Charles P. Knleberger, Amsteram: orth- Hollan, pp Wng, I., (2004) Computable General Equlbrum Moels an Ther Use n Economy-We Polcy Analyss. MIT Jont Program on the cence an Polcy of Global Change. Techncal ote 6. 60

69 Chapter 2. Appences Appenx A The moel specfe n Chapter 2 brngs us very close to the stanar CGE verson of the xt-tgltz moel of monopolstc competton as n Markusen (2002) (Moel M62- mcp.gms). In fact, the only fferences are n the mcro-consstency matrx to whch the moel s calbrate, the values of the parameters, the a valorem tax an the economc nterpretaton. Inee, the representatve consumer coul be emanng the varetes an therefore usng a two-stage bugetng process. The obtane eman woul be the same as n our moel specfcaton. The same moel structure coul sometmes be use for nterpretaton of fferent economc problems. Ths enables us to obtan new nsghts nto the exstng approaches. In the coe of the moel, the equatons for the monopolstcally compettve sector are gven at the varety level. Therefore the correctness of our program coe coul be checke by transformng the equatons of our moel to the varety level an comparng t to the stanar coe. After all the necessary transformatons, the coes are entcal. The moel n Markusen (2002) coul be mae structurally smlar to the smplfe verson by: () nclung an a valorem tax (2) ang an external sector. The tax (t) s taken to be 0.25 as n the base moel. It s also assume that the economy s exportng a perfectly compettve goo an mportng an IR goo n the benchmark. mlarly to our moel, the value of exports of one goo equals the value of mports of the other goo: TEY TMX 0. The results of the mofe moel n both the orgnal an our moel notaton of varables are gven n the frst column of Table 2.A. Table 2.A. Percentage change from the benchmark after tax cut (t = 0) Our moel Orgnal varable Intal case 0 8 equvalent notaton ( ) ( ) x X % -38% -28% Z Y 40% -00% -22% W W -% 3% 5% -49% 27% 57% PX E 0% 0% 0% px PX -6% 8% 2% PZ PY 0% 0% 0% w PZ -0% 6% 39% v PW 7% 3% -8% PW PU 0% 0% 0% CO CO -% 3% 5% TEZ-TMZ TEY-TMY 45% -49% -39% (TMX- TMX-TEX 309% -338% -00% TEX) e PFX 0% 0% 0% X X c -52% 33% 26% 6

70 Chapter 2. Appences Clearly, the seconomes of scale ajustment omnates results. mlarly to the ntal moel, two key parameters, namely the fference n the sklle labor ntenstes an the elastcty of substtuton between varetes, shoul be mofe to make the factor market crowng ajustment stronger than the seconomes of scale ajustment. Consequently, the results presente n the secon an thr columns of Table 2.A contan the stable result obtane by ncreasng the value of the substtuton elastcty or the fference n the sklle labor ntenstes. The mcro-consstency matrx use to obtan ths result s presente n Table 2.2A. Even though the magntues of the effects n Table 2.A are completely fferent from the ones presente n Table 2.2 ue to the fferent ata set, the rectons of change of all the varables are the same an coul be fully explane by Proposton n secton 3. Table 2.2A. Mcro-consstency matrx of the mofe M62 moel Proucton Proucton sectors Consumers Markets X c X Y TEY TMX U CO PX E PY PFX 0-0 PU PW PZ A valorem tax (t) Fxe cost Appenx B: ynamc tme paths of the varables n the EG specfcaton Fgure B. ynamc tme path of the output of goo X T X 62

71 Chapter 2. Appences Fgure 2B. ynamc tme path of the output of goo Z T Z Fgure 3B. ynamc tme path of the number of frms T Fgure 4B. ynamc tme path of the welfare T W Fgure 5B. ynamc tme path of q Theta T

72 Chapter 2. Appences Fgure 6B. ynamc tme path of the usklle laboe wage. 0.9 v T Fgure 7B. ynamc tme path of the consumer expenture T CO Fgure 8B. ynamc tme path of the frm level prce n nustry X px T Appenx C: Program coe $TITLE On Equlbrum tablty n the Moel wth Monopolstc Competton * Moel wth AVALOREM TAX * HERE MC A FC HAVE THE AME IPUT COMPOITE * Z-perfectly compettve sector * X-ector wth mperfect competton * Elastcty of substtuton between frm types: 5 ET THETA /ter*ter30/ ; PARAMETER IGMA Elastcty of substtuton between varetes n the IR sector, T Output tax on servce prover, FC Calbraton parameter, FC Calbraton parameter, PEZ Export prce of goo Z, PMZ Import prce of goo Z, PEX Export prce of goo XF, PMX Import prce of goo XF px0 Frm level (varety) prce n the benchmark, 64

73 Chapter 2. Appences x0 Frm level (varety) output n the benchmark, pxot Frm level (varety) prce after tax, xot Frm level (varety) output after tax, ALPHA The share of goo X n the consumer expenture, BETA kll ntensty of the CR sector, GAMMA kll ntensty of the IR sector, P Relatve worl after tax prce of goo X, _EOW Total Enowment of klle Labor, L_EOW Total Enowment of Unsklle Labor, Q(THETA) Relatve output of goo X; IGMA =5; T=0.25; ALPHA=0.725; GAMMA=0.75; BETA=0.3; FC= 0**(/(-IGMA)); _EOW=222.5; L_EOW=2327.5; PEZ=; PMZ= 000/999; PEX=999/000; PMX=; POITIVE VARIABLE W! Consumpton (statc welfare) level Z! Actvty level for sector Z X! Actvty level for sector X! umber of servce frms TMZ! Actvty level for mports of Z TEZ! Actvty level for exports of Z TMX! Actvty level for mports of Y TEX! Actvty level for mports of Y PW! Prce nex for consumpton PZ! Prce nex for commoty Z PX! Prce nex for commoty Z RL! Return to unsklle labor L (all other factors composte) R! Return to sklle labor (sklle labor) E! Prce nex for foregn exchange CO! Income level for consumer CO; EQUATIO PrceW Zero proft conton for welfare PrceZ Zero proft conton for Z sector PrceX Zero proft conton for X E_TEX Zero proft conton for exports of X E_TMZ Zero proft conton for mports of Z E_TEZ Zero proft conton for exports of Z E_TMX Zero proft conton for mports of X W Market clearng conton for W Z Market clearng conton for Z X Market clearng conton for X PFX Trae balance equaton RL Market clearng conton for L (all other factors composte) R Market clearng conton for (sklle labor) An equaton to etrmne ICO Income balance for representatve consumer; * Zero proft contons: PrceX.. R**GAMMA*RL**(-GAMMA)=G=**(/(IGMA-))*PX*(-T)*(-(/IGMA))*FC; PrceZ.. R**BETA*RL**(-BETA)=G=PZ; PrceW.. PZ**ALPHA*PX**(-ALPHA)=G=PW; 65

74 Chapter 2. Appences E_TEZ.. PZ=G=E*PEZ; E_TMX.. E*PMX=G=PX; E_TEX.. PX=G=E*PEX; E_TMZ.. E*PMZ=G=PZ;.. R**GAMMA*RL**(-GAMMA)*8**0.875=G=(/IGMA)*(PX)*(-T)*X*400; * Market clearng contons: W.. Z.. X.. RL.. R.. W*3600=G=CO/PW; 3200*Z+299*TMZ=G=300*TEZ+PZ**(ALPHA-)*PX**(-ALPHA)*W*2900; 400*X+300*TMX =G= 299*TEX+PZ**ALPHA*PX**(-ALPHA)*W*700; L_EOW=G= (-BETA)*3200*Z*R**BETA*RL**(-BETA) +(-GAMMA)*X***(/(-IGMA))*270*R**GAMMA*RL**(-GAMMA)/FC +(-GAMMA)**8*R**GAMMA*RL**(-GAMMA); _EOW=G= BETA*Z*3200*RL**(-BETA)*R**(BETA-) +GAMMA*X***(/(-IGMA))*270*RL**(-GAMMA)*R**(GAMMA-)/FC +GAMMA**8*RL**(-GAMMA)*R**(GAMMA-); PFX.. TEX*PEX+TEZ*PEZ=G=TMZ*PMZ+TMX*PMX; *Income balance conton: ICO.. CO=E= L_EOW*RL+_EOW*R+T*X*PX*400; Moel MoTAB /PrceW.W, PrceZ.Z, PrceX.X, E_TEX.TEX, E_TMZ.TMZ, E_TEZ.TEZ, E_TMX.TMX, W.PW, Z.PZ, X.PX, RL.RL, R.R,., ICO.CO, PFX.E/; *Assgnng ntal values to enogenous varables W.L=; Z.L=; X.L=;.L=0; TEX.L = 0; TMZ.L = 0; TEZ.L = ; TMX.L =; PW.L=; PZ.L=; PX.L=; R.L=;RL.L=; CO.L=3600; FC= /(0**(/(IGMA-))*(-T)*(-(/IGMA))); px0=px.l*.l**(/(igma-)); x0=x.l/(.l*.l**(/(igma-))); *umerare E.FX=; OLVE MoTAB UIG MCP; loop (THETA, P= *(OR(THETA))/(CAR(THETA)); T=-P; OLVE MoTAB UIG MCP; Q(THETA)=X.L/Z.L; ); execute_unloa 'upply_prce.gx', Q ; fle f /supply_prce.gch/; put f; $onput [CHART] GXFILE=upply_prce.gx TITLE=Graph of the supply prce [ERIE] YMBOL=Q TYPE=lne $offput 66

75 Chapter 3 FI n strbuton ervces an Trae versus Investment Trae-Off In ths chapter, I construct a basc analytcal moel to stuy ncentves to nvest n strbuton servces by a vertcally ntegrate multnatonal frm. Therefore, a framework wth alternatve optons of foregn market penetraton such as exportng an FI n manufacturng are consere. There s a market seekng motve to nvest n strbuton servces an there s a cost orente motve to nvest n manufacturng. The parameter values are assume to match the stylze facts on nvestment n strbuton. The two country north-south moel prouces several analytcal results. In partcular, market sze n the host country, prouctvtes of proucng fferentate goos, substtutablty an the host country strbuton margn ncrease the ncentve for FI n strbuton. On the other han, transportaton cost an wage rates n both the host an the home country of the multnatonal ecrease strbuton FI. Moreover, epenng on the level of trae openness, lberalzaton of trae n goos may both ncrease an ecrease the level of nvestment n strbuton servces. 3.. Introucton The growth of multnatonal enterprse actvty n the form of FI (Foregn rect Investment) has been rsng at a faster rate than trae flows between countres. The exstng emprcal evence on etermnants of a frm s ecson to nvest abroa assgns a greater sgnfcance to horzontal rather than vertcal FI (avarett an Venables, 2004). Ths s apparently the case, snce most of nvestments stll take place between evelope countres where market seekng motves are more prevalent. On the other han, factor prce fferentals, whch are consere to be a central cause of vertcal FI, woul be nsgnfcant to explan nvestments n-between hgh ncome countres. However, the mportance of vertcal FI s ncreasng. Moreover, t s ffcult to sentangle the effects of factor prce fferentals (Hanson et al., 200). Therefore, the sgnfcance of vertcal FI mght be unerestmate an t s better to stuy vertcal FI n the regonal an sectoral specfcaton where t shoul have a clearer pattern. In ths chapter, we construct a theoretcal framework that stngushes between nvestments n strbuton servces an manufacturng. In ths way we allocate a specfc type of nvestment to a specfc sector. amely, FI n strbuton servces s market seekng an FI n manufacturng s rven by factor cost fferentals. Consequently, the am of the chapter s to create a relatvely smple analytcal framework that woul generate emprcally testable prectons about the etermnants of FI n strbuton. In aton, there are almost no theoretcal approaches to FI etermnants to my knowlege that woul explctly scrmnate between FI n strbuton an n manufacturng. Why strbuton servces? Accorng to the WTO classfcaton, the strbuton servces sector comprses wholesale an retal trae, commsson agent s servces an franchsng. In other wors, those are the servces that consttute an essental lnk between consumers an proucers both wthn an across borers. There s a growng tenency towars more vertcal ntegraton wthn the strbuton sector tself. Improvements n nformaton technologes an 67

76 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off logstc servces mae t possble for large retalers to ntegrate wholesale functons as well. Thus, ths chapter focuses on the strbuton sector n general. The sector accounts for up to 50% of the prce pa by consumers (Plat, 997). Trae n two major components of strbuton servces: nternatonal wholesale an retal trae occurs manly through commercal presence or moe 3 of GAT (General Agreement on Trae n ervces). Moreover, n sectoral terms, the composton of FI n general has shfte towars servces. The strbuton sector s connecte to manufacturng n several ways. The retal nustry, for example, sets prouct stanars, shares nformaton on prouct promoton an consumer preferences wth the manufacturng nustry. Moreover, the proft of the manufacturng sector epens on promotonal servces prove by retalers such as pre-sale nformaton, avertsng, crets, free elvery an so on (Trole, 988). That establshes potental benefts from vertcal ntegraton wth the retal sector. Furthermore, many manufacturng frms have ether ther own strbuton chans (see cars or luxury goos) or have bnng contracts wth wholesale traers to control the marketng of ther proucts. A ratonal queston that the reaer may have at ths pont s: Is there any evence for multnatonal frms to vew the two types of nvestments as alternatves? Hanson et al. (200) stuy the ata on prmary an seconary nustres of the foregn afflates of U base multnatonal frms whose parent prmary nustry s manufacturng. They fn that the secon major nustry of the foregn afflate sales of those frms s wholesale trae. Moreover, controllng for the sze of the frm they fn that most of the frms chose to nvest n ether manufacturng or wholesale trae an rarely n both xv. Consequently, the authors argue that FI shoul be fferentate as proucton orente an strbuton orente. The subsequent feature of the moel that allows nvestments n manufacturng an/or n strbuton by the frm whch s vertcally ntegrate across both of those sectors omestcally s motvate by ths emprcal pattern. At the same tme, there s a lot of lterature on substtutablty an complementarty between FI an trae n general. On the theoretcal front, theores on horzontal FI prect a negatve relatonshp between FI an trae xv an theores on vertcal FI prect a postve relatonshp. However, emprcal evence ncates that t s more complex than that. Pontes (2004) crtczes the stanar theores of FI for assumng a monotonc relatonshp between FI an trae. He then obtans a non-monotonc relatonshp, n a moel wth vertcally relate ntermeate an fnal stages of proucton. There s also a growng boy of emprcal research on substtutablty/complementarty between FI an trae. Most of t supports the complementarty relatonshp. For example, wensson (2004) fns that complementarty s prevalent at the hgh levels of ata aggregaton. However, at the lower levels of aggregaton corresponng to the prouct lnes substtutablty s observe. Ths proves an explanaton on the screpancy between theoretcal moels that usually assume substtutablty an exstng emprcal evence. Hea an Res (200) stuy the relatonshp between exports an strbuton FI. They fn sgnfcant evence supportng the assumpton of ther theoretcal moel an, as many other emprcal stues, the complementarty relatonshp between FI n manufacturng an exports. xv Here controllng for the sze of the frm means that the effect oes not appear only because foregn afflates of relatvely smaller frms ten to specalze n one sector. xv Also takng nto account subsequent trae wth ntermeates. 68

77 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Moreover, the postve relatonshp between FI n strbuton an exports was foun to be stronger than the one for manufacturng, as t was expecte. On the other han, oras et al. (2007) fn that FI n the strbuton servces sector woul rase mports from the host to the home country of the multnatonal frm. In terms of the polcy relevance of the gven ssue, t shoul be note that there s a wesprea vew n evelopng countres that FI n strbuton servces yels lttle evelopment mpact an t s not crucal for growth (Palmae et al., 2004). At the same tme, evelopng countres make a lot of effort to attract FI n manufacturng that coul be brought by effcency seekng multnatonal frms. As a matter of fact, n terms of the level of FI, manufacturng stll remans to be the poltcally most preferre sector n evelopng economes. It s also feare that more nvestments n strbuton coul create urban sprawl an push smaller an less effcent retalers out of the market, whch n turn coul potentally brng more unemployment. As a matter of fact, there s an ongong ebate whether to allow FI n the retal sector or not n several evelopng countres lke Ina an Inonesa. On the other han, the restrctons on strbuton coul hner the effcency of the upstream nustres an consequently brng hgher busness costs. Restrctons coul also lea to rent-creaton by ncumbent frms. Inee, Kalrajan (2000) fns that restrctons are prmarly cost-creatng. Market structure of the strbuton sector vares from country to country. However, there s a general tren for hgher market concentraton. For example n the retal sector, ths coul be effcency mprovng ue to the sgnfcant economes of scale (oras et al., 2007). In aton to that, retalng s characterze by a eclne n the total number of outlets an at the same tme by an ncrease n the number of large megastores. The mpact of the market structure n strbuton on the proft of the vertcally ntegrate frm an on the upstream frm s consere n the last secton. Another stylze fact about retalng s that the number of offere varetes of proucts has ncrease over tme (Eckel, 2009). Apparently, those changes are the result of ncreasng globalzaton an freer trae. The above mentone stylze facts are ncorporate n the moel constructe n ths chapter when trae openness brngs more FI n strbuton. There are several stues relate to the nteracton between trae n strbuton servces, market concentraton an trae n goos. In partcular, Francos an Wooton (2007) fn that a rsng concentraton n strbuton servces coul be an effectve barrer for trae n goos. As for the goos trae, Ishkawa (2008) shows that trae lberalzaton n goos, f not accompane by FI n strbuton servces, coul hurt omestc consumers an lea to the an overall welfare fall. The rest of the chapter s organze as follows: the next secton presents an outlne of the theoretcal moel. In the thr secton we present the parameter assumptons an the corresponng analytcal results. The fourth secton presents numercal results an the nonmonotonc relatonshp between trae costs an FI n strbuton an the ffth secton conclues. 69

78 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off 3.2. Moel settng A two country moel s consere. It s a stylze fact that FIs orgnate preomnantly from evelope countres (avarett an Venables, 2004). Therefore, I assume that the multnatonal frm s locate n a evelope country an s enote by north. mlarly, a evelopng country receves FI an s enote by south. It s assume that each economy s compose of a monopolstc sector of two fferentate traable goos an a compettve numerare sector. Intally, only one of the fferentate goos coul be prouce n each country. The vertcally ntegrate monopolst maxmzes profts by servng two segmente markets. In other wors, n aton to the local market t coul serve the foregn market through nvestments or trae. We assume a quaratc utlty functon: U x y z z x y x xy y 2 2 (,, ) x y ( x 2 y ) / 2 (3.2.) Here z s a numerare goo an x, y are two fferentate goos trae by a multnatonal frm. The functonal form of the utlty functon enables us to conuct a partal equlbrum analyss of the fferentate goos only. Ths s because markets for numerare an non-numerare goos are separate whch leas to no ncome effects on those goos. Consequently, the erve eman functons wll be lnear an wll look as follows, for example, for the evelope country: x a b P c P y a b P c P c c x x y x x c c y y x y y (3.2.2) j j Here a ; b ; ;,, {, } 2 c 2 x cy j j x y 2 j j j Let us assume a symmetrc case wth a a : a an b b : b. Moreover, let us x y x y assume that c c. Ths s not a loss of generalty because unts can be chosen so that x y both coeffcents equal one. Then, (3.2.2) wll smplfy to: x a P b P y a P b P c c x y c c y x (3.2.3) Ths form enables us to gve meanng to the remanng parameters: a 0 s the market sze n the north an b [0,] represents the substtutablty between the two goos. mlarly for the evelopng country: x a P b P y a P b P c c x y c c y x (3.2.4) The Y-goo s prouce n the south an the X-goo s prouce n the north. We coul also thnk of those goos to be compostes of the foregn an omestc goos of a multnatonal 70

79 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off frm. There s only one factor of proucton. Thus, the proucton functon n the manufacturng nustry Y n the south, for nstance, coul be represente as xv : y L (3.2.5) Here enotes prouctvty. mlarly n the north: x L (3.2.6) The multnatonal coul also shft the proucton of ts goo to the evelopng country: x L (3.2.7) In contrast to proucer servces, where sklle labor ntensty s an mportant stngushng factor, strbuton servces employ many low sklle workers. fferently from FI n manufacturng, FI n strbuton s expecte to brng lttle mpact on employment n the host country. On the contrary, t may actually ecrease labor eman. Therefore, the corresponng varable labor costs may both ncrease or ecrease after FI n strbuton. Consequently, n ths settng, we neglect varable costs of strbuton. They coul also be thought of to be partally nclue n varable proucton costs. The margnal costs of proucton wll be later chosen to set the unts of measurement n the reference equlbrum. On the other han, there wll be stnct fxe costs of FI n strbuton. nce no portfolo nvestments are consere n ths moel, the wors FI an nvestment are use nterchangeably. There wll be four scenaros epenng on the ecson of the multnatonal frm to serve the foregn market through exports, to nvest n manufacturng, to nvest n strbuton servces or to nvest n both. In what follows, I woul lke to conser all four cases or moes of supplyng the southern market separately Case : o nvestments In ths case the evelope country exports the local goo, mports the foregn goo an no nvestments take place. The multnatonal frm wll maxmze ts proft at home an abroa by optmally choosng the prces t charges for both goos. In partcular, the omestc proft s gven as follows: ( P P ) y ( P P ) x (3.2.8) c p c p y y x x The proucers prces are obtane routnely from corresponng maxmzaton contons: p Py w p w (3.2.9) Px (3.2.0) t Here w, w enote wages n the north an south corresponngly, s a transportaton cost an t s an a valorem sales tax mpose by the north on the mportng frm. The multnatonal frm whch has a full monopoly over goo X wll also export t to the south an maxmze the proft by chargng proucer prce P : xv p Px w ( ) x (3.2.) t p x xv I coul have chosen a form wth plant-level economes of scale. However, n the settng where proucton takes place ether n two countres or n the south only, that woul be equvalent to ang an atonal ncentve for FI n manufacturng. The effect wll be consstent wth theoretcal an emprcal prectons of plant-level economes of scale on FI n manufacturng. It won t brng any sgnfcant changes to the man outcomes of the moel. xv The upper nex s use to ncate whether we use proucer s prce (p) or consumer s prce (c). 7

80 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Here t s an a valorem tarff mpose by the south. I assume that there are many small strbuton outlets n the south an the strbuton margn s fxe at the level K. Hence, consumer prces wll be etermne as follows: c c p Py w K (3.2.2) Px Px K (3.2.3) On the other han, monopoly n the strbuton sector woul result n a problem of ouble margnalzaton wth very hgh prces an very small outputs of goos. Then, the combne proft of the strbuton an manufacturng sectors wll be smaller than the vertcally ntegrate proft an consumers wll be worse off Case 2: FI n manufacturng In ths case, the multnatonal frm shfts the proucton of the northern country goo to the south. Accorngly, both goos wll be prouce n the south an ncur transportaton costs an taxes. In ths case, no goos wll be prouce n the north an the labor of the northern country coul be thought to move to heaquarter actvtes even though t s not explctly moele. Ths assumpton s rather strong an t constrans labor enowment of the south to be large relatve to the labor enowment of the north n orer to prouce feasble values. The proucton functon s represente by (3.2.7), wth hgher prouctvty n the north: w w.however, t woul also be ntutve to have. In other wors, even though t s more effcent to prouce goo X n the north, the unt cost of proucng X s lower n the south. That ncorporates the vertcal motve to nvest n the south. Atonally, there wll also be an mplct horzontal motve snce the multnatonal frm supples to the host market as well. As n the case before, multnatonal frm wll maxmze ts profts by chargng consumer prces for both goos omestcally an the proucer prce for the X-goo n the south: 2 c p c p c w ( Py Py ) y ( Px Px ) x ( Py ( t ) ) y c w ( Px ( t ) ) x (3.2.4) 2 p w ( Px ) x FM (3.2.5) Here FM s a fxe cost of FI n manufacturng. Labor s assume to be mmoble between countres. Therefore, only the resources of the southern country wll be use n the fnal proucton Case 3: FI n strbuton servces In ths case, the multnatonal frm nvests only n strbuton servces. I assume that t takes over the less effcent strbuton sector of the south an becomes the only strbutor n the south just as t s alreay n the north. Ths effect s justfe by the cost-preaton conseratons scusse below. The multnatonal frm ncurs no strbuton costs. The profts n both markets are etermne as follows: 72

81 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off w w ( P ( t ) ) y ( P ) x (3.2.6) 3 c c y x 3 c p c p c w ( Py Py ) y ( Px Px ) x F ( Py ) y c w ( Px ( t ) ) x F (3.2.7) Here F s a fxe cost of FI n strbuton servces. It conssts partally of the cost of acquston of the southern strbuton sector an the cost of erectng no entry barrers as well as the fxe costs of operatng n a foregn market. The strbutor sources the Y-goo locally an the X-goo from ts own home country. Apparently, there s a horzontal motve to nvest n strbuton servces Case 4: FI n both manufacturng an strbuton servces The profts n the north an n the south wll be gven as follows: 4 2 (3.2.8) 4 c p c p c w ( Py Py ) y ( Px Px ) x ( FM F ) ( Py ) y c w ( Px ) x ( FM F ) (3.2.9) Here s a measure of complementarty between two types of nvestments. Thus, profts n the north change only wth nvestments n manufacturng. The schematc representaton of the strateges of the multnatonal frm s gven n Appenx A. Exstng nustral organzaton lterature, n most cases smlar to the assumptons of the moel, supports the concept of full forwar vertcal ntegraton by a monopolst frm an the corresponng takeover of ownstream nustry. In other wors, n the case wthout economes of scale an wth fxe-proporton ownstream proucton technology, the upstream monopolst woul be able to eject the competng ownstream frms (frnge frms). Ths effect occurs ue to the strategc cost-preaton conseratons when a omnant monopolst rves compettors out of the ownstream market by strategcally ncreasng prces of the nputs t supples. Bonroy an Larue (2007) fn a partal forwar ntegraton n the moel wth ecreasng returns to scale n the ownstream nustry an ncreasng returns to scale n the upstream nustry. In the present moel, one unt of manufacturng nput s requre to prouce one unt of output n strbuton. That represents a fxe-proporton technology. The upstream frm supples only one goo, the prce of whch, uner average values of substtutablty, coul sgnfcantly ncrease the costs of potental frnge frms. Therefore, n the gven moel settng, the assumpton of full forwar vertcal ntegraton after FI n strbuton s justfe. Returnng to the moel settng, the total proft that the multnatonal frm wll obtan n each moe s just the sum of the profts n the north an n the south. The proft functons epenng on the parameters of the moel coul be obtane after solvng the corresponng proft maxmzaton problems: 73

82 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off ( t, t, w, w,, K,,, b) max ( ) c c p Px, Py, Px ( t, w,, K,,, b, FM ) max ( ) c c p Px, Py, Px (3.2.20) ( t, t, w, w,,,, b, F) max ( ) c c c c Px, Py, Px, Py ( t, w,,,, b, FM, F, ) max ( ) c c c c Px, Py, Px, Py The optmzaton problem coul be smplfe f we note that max ( ) max max, 2 c c p p c c for Px, Py, Px Px Px, Py (3.2.2) max ( ) max max 3, 4 c c c c c c c c for Px, Py, Px, Py Px, Py Px, Py Any statonary ponts obtane from optmzaton wll always be a maxmum because of the concavty of profts n (3.2.2). For the profts n (3.2.2) to be concave, the corresponng Hessan matrx shoul be negatve sem-efnte at the statonary pont. The sgn efnteness of the Hessan matrx s equvalent to the sgn efnteness of the corresponng quaratc form at the statonary pont. The negatve efnteness of the quaratc form coul be emonstrate as follows: 2 2 0, for, 2 2 P P t x , 4( ) 0, 3, 4 2 c 2 c 2 c 2 c b for 2 c c Px Py Px P y Px P y , 4( ) 0, 2,3, 4 2 c 2 c 2 c 2 c b for 2 c c Px Py Px P y Px P y Let n (3.2.22) enote the feasble set of the parameters of moe. In other wors, n that set, prces, quanttes an profts are greater or equal than zero. * * * * { parameters x ( P ) 0, x ( P ) 0, y ( P ) 0, y ( P ) 0, P 0, P 0, P 0, P 0, ( P ) 0} (3.2.22) * * * * * x y x y The number of the parameters efne n wll change n fferent specfcatons. We efne as the feasble set of parameter values of our moel to be an ntersecton of the feasble sets of each moe: Analytcal results etermnants of FI n strbuton In ths subsecton we restrct our analyss to ncentves to nvest n strbuton both wth an wthout FI n manufacturng. nce profts n the north are the same n moes an 3 as well 74

83 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off as n moes 2 an 4, we coul restrct our analyss only to profts n the south. However, northern profts wll alreay nee to be taken nto account n the next subsecton. In orer to stuy the mpact of the parameters on the ecson to nvest n strbuton, we efne the followng functons: , M (3.3.) Hence, the ervatves of those functons wll prove us wth an unerlyng ntuton of the mpact of the parameters. In partcular, a postve value of the ervatve means more strbuton FI. The ncentve to nvest n strbuton s stue both uner a prohbtvely hgh cost of manufacturng (0) an n the case when manufacturng nvestment has alreay taken place (M). Accorng to the envelope theorem, the mpact of the changes n parameter values woul ncorporate the mpact from the changes n maxmzer varables an therefore a partal ervatve of the proft functon woul exactly reflect the change n the optmally chosen value functon. Moreover, the proft fferences n (3.3.) coul be reuce to: , M Ths s because profts n the north are the same for moes an 3 as well as for moes 2 an 4 as coul be seen from (3.2.2). Therefore country specfc nexes for the market sze an sustanablty coul be roppe: a a, b b b. Let me enote the feasblty sets corresponng to 0 an M as 0F an MF: 0F 3 ; MF 2 4 ; Hence, the followng proposton coul be prove: Proposton : Let us assume the case where FI n manufacturng takes place, all of the parameters belong to the feasblty set MF an strbuton markups are nonnegatve. Then, the market sze n the south (a), fxe cost of manufacturng (FM), prouctvtes of proucng both fferentate goos n the south (, ), southern market neffcency n strbuton ( K ) woul ncrease the ncentve to nvest n strbuton an the southern wage ( w ) an a fxe cost of strbuton (F) woul ecrease t. Moreover, f a w K (3.3.2) substtutablty (b) stmulates FI n strbuton. However f b a w K, then b (3.3.3) s a suffcent conton for a postve mpact of b substtutablty on ncentve to nvest n strbuton. The last conton consers the case where market sze n the south s suffcently small for the markup that the multnatonal frm earns on the Y goo to be less than the omestc strbuton margn: 0 a w K b. Then, the prouctvty of the locally prouce goo shoul not be much less than the prouctvty of the goo prouce by the multnatonal n the south, takng nto account the substtutablty between the two goos. Otherwse, an ncrease n the substtutablty between the fferentate goos may not lea to FI n strbuton. A smlar proposton coul be formulate for the case uner a prohbtvely hgh fxe cost of manufacturng. 75

84 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Proposton 2: Let us assume the case wth no FI n manufacturng, all parameters belongng to the feasblty set 0F an nonnegatve strbuton markups. Then, the market sze n the south (a), prouctvtes of proucng fferentate goos (, ), southern market neffcency n strbuton ( K ) woul ncrease the ncentve to nvest n strbuton an fxe cost of strbuton (F), transportaton cost ( ), southern wage ( w ) an northern w w wage ( w ) woul ecrease t. Moreover, f a ( b) K (3.3.4) t substtutablty (b) has a postve mpact on ncentve to nvest n strbuton. In orer to prove propostons an 2 we nee to show that the followng contons hol M M M M M M M M 0, 0, 0, 0, 0, 0, 0 an 0 a FM K w F b a w a w for K K b ; b b for Proposton an the followng contons hol , 0, 0, 0, 0, 0, 0, 0 an 0 a K F w w b f (3.3.4) hols; for Proposton 2. The proofs of both propostons are gven n Appences B an C. Many parameters n Propostons an 2 rectly nfluence the quanttes of goos. Furthermore, t s possble to show that the quantty of the goo that the multnatonal sources from the home country (X goo) wll be larger wth FI n strbuton than wthout t uner both specfcatons: * * * * w w x0 x ( Px 3, Py 3) x ( Px, Py ) a b ( t ) 2 bw w a ( b) K ( t ) ( b) K t 0 (3.3.5) 2 2 * * * * w w bw w xm x ( Px 4, Py 4) x ( Px 2, Py 2) a b a ( b) K 2 2 ( b) K 0 (3.3.6) 2 The mpact on the output of the locally prouce goo (Y-goo) s less obvous. However, we coul emonstrate that the prce of that goo wll be larger or smaller wth FI n strbuton than wthout epenng on whether the monopoly markup s hgher or smaller than the local strbuton margn K : a w w a w P P P P P P K K 2 b 2 b * * * * y0 y3 y ym y 4 y2 monopoly markup (3.3.7) 76

85 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off In the case when the strbuton markup of the multnatonal frm s smaller than K, the prce of the Y-goo ecreases after FI n strbuton ( P 0 0, P 0 ) an t coul be shown usng (3.2.4) that the prce of the X goo also ecreases after FI n strbuton: P bp x or P bp x 0. x0 y0 0 0 xm ym M On the other han, f the strbuton markup s hgher than K, an the prce of the Y-goo s larger wth FI n strbuton than wthout t ( P 0 0, P 0 ), t coul be shown that the output of that goo s smaller wth FI n strbuton than wthout: y0 0, y M 0. Ths coul be prove by usng the fference form of (3.2.4) an conserng both cases wth ncreasng an ecreasng prce of the X-goo after FI n strbuton: f P x0 0 then y0 Py 0 bp x 0 0. If P x 0 0 then 0 0 y y y0 Py 0 bp x 0 bpy 0 Px 0 ( x0 ) 0. o n ether case y y M mlarly t coul be shown that: 0. In other wors, the followng proposton has been prove: Proposton 3: Let us assume that all the parameters belong to the feasblty set an strbuton markups are nonnegatve. Then, the output of the goo that the multnatonal frm sources from ts home country s hgher wth FI n strbuton than wthout. Moreover, f the strbuton markup of the locally source goo (Y) s lower than the local strbuton cost ( K ), enote as case (I), the fnal prces of both goos sol by the multnatonal frm wll eclne after FI n strbuton. On the other han, f the strbuton markup of the locally source goo s hgher than the local strbuton cost, enote as case (II), the fnal prce of the local goo wll ncrease an the quantty of t wll eclne after FI n strbuton: a w ( I ) K 0 Py 0 0, PyM 0, Px 0 0, PxM 0; 2 ( b) a w ( II) K 0 Py 0 0, PyM 0, y0 0, ym 0; 2 ( b) Ths proposton s nepenent of the fxe cost of manufacturng an wll help us nterpret the results from Propostons an 2: ) Market sze (a): A hgher market sze n the south woul result n hgher eman from the south an consequently more potental proft after nvestment n strbuton. 2) Prouctvtes (,, ) : A hgher prouctvty of proucng ether goo both n the north an n the south woul make proucton an trae less costly. That woul stmulate FI n strbuton. 3) Transportaton cost ( ) : Obvously, n the case where all proucton takes place n the south, transportaton costs are ncurre only to the north an as a result o not matter. Otherwse, the multnatonal frm chooses the moe wth fewer exports to the south an the corresponngly smaller per unt cost of transportng t. 4) outhern strbuton margn ( K ) : Hgher southern market neffcency n strbuton woul wen the gap between the level of exports to the south wth an wthout FI n ym ym 77

86 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off strbuton. Therefore, the amount of potental rent that the monopolst coul extract after nvestng n strbuton rses wth K. Ths leas to the postve sgn wth respect to K. 5) outhern wage ( w ) : Hgher costs of proucton n the south woul result n lower monopoly rent after strbuton FI. Although the negatve effect s unambguous n moes 2 an 4, n the case wthout manufacturng FI there wll also be a smaller counter effect. amely, the hgher cost of proucton n the south woul rase the prce of the Y-goo an consequently ncrease the eman for the X-goo. Ths shoul stmulate strbuton FI snce t brngs more exports of the X-goo. 6) orthern wage ( w ) : Here, as the cost of proucng the X-goo an accorngly ts prce ncrease, the strbuton markup wll shrnk, the moe wth no FI n strbuton an less exports wll be chosen. Even though labor s not employe n strbuton n ths moel, hgher wages n both the host an the home country of the multnatonal frm woul result n less FI n strbuton. 7) ubsttutablty (b): More substtutablty, also entalng hgher consumer flexblty n choosng between the two goos, woul restrct monopoly power n the moes where the multnatonal supples only one goo to the southern market. Ths shoul stmulate FI n strbuton. One of the ncentves for FI n strbuton s elmnatng the neffcency assocate wth unerconsumpton of the X-goo prouce by the monopoly relatve to the compettvely prouce Y-goo. However, ths effect coul be small or even reverse when the prouctvty of the Y-goo s small relatve to the prouctvty of X-goo for a gven value of the substtutablty between the goos. Then, consumers woul have less ncentve to substtute the X-goo for the more expensve Y-goo. Consequently, when the market sze n the south s so small that conton (3.3.2) oes not hol, the postve mpact of FI n strbuton s also small. In that case, for the postve mpact of the substtutablty on the ncentve to nvest n strbuton t s suffcent f the rato of corresponng prouctvtes satsfes conton (3.3.3). In the case where only moes wth an wthout strbuton are compare, the segmente markets are nepenent of each other. Therefore, the northern market sze as well as the tarff mpose by the north woul have no mpact on the ncentve to nvest n strbuton. The mpacts of the fxe costs are trval. The obtane results coul be contraste wth exstng emprcal stues. In partcular, Hanson et al. (200) fn that the share of FI n the wholesale trae sector n the total of wholesale an manufacturng FI s ecreasng n the host country GP. Ths result s nconsstent wth the horzontal theores of FI, especally when apple to strbuton servces. Boatman (2007) uses agency theory conseratons to stuy the etermnants of FI n strbuton. In that stuy, frms choose between supplyng the foregn market by nvestng n strbuton servces or by exportng. Both Hanson et al. (200) an Boatman (2007) fn a postve an sgnfcant relatonshp between per capta GP n the host country an FI n strbuton. We coul raw parallels to our stuy because per capta GP coul be use as an alternatve measure of market sze. 78

87 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off etermnants of FI n manufacturng The moel has not been constructe to prove a comprehensve framework to fn all of the etermnants of FI n manufacturng. However, the mpact of some parameters coul also be shown analytcally an atonal assumptons on the values of the parameters coul be erve. Those assumptons woul be useful n the numercal analyss. mlarly to the case for strbuton, let us efne the followng functons: , (3.3.8) Then, the followng Proposton on the etermnants of FI n manufacturng coul be prove: Proposton 4: Let us assume that all parameters belong to the feasblty set, strbuton w w markups are nonnegatve an the followng conton hols: ( tn ) (3.3.9) Then, the market sze n the south ( a ) an n the north ( a ),substtutablty (b),the northern wage ( w ), prouctvty of proucng an outsource goo n the south ( ), an southern tarff ( t ) woul ncrease the ncentve to nvest n manufacturng. On the other han, southern market neffcency n strbuton ( K ), prouctvty of proucng an outsource goo n the north ( ), prouctvty of proucng a local goo n the south ( ) woul ecrease the ncentve to nvest n manufacturng. Moreover, f 2 b (3.3.0), the southern wage ( w ) an the northern tarff ( t ) woul also ecrease the ncentve to nvest n manufacturng. The proof of Proposton 4 s gven n Appenx. Proposton 4 hols n both scenaros wth an wthout FI n strbuton. Furthermore, the fference n profts n the north ( ) s the same rrespectve of FI n strbuton an coul be wrtten as follows: w w w w a b ( tn) a ( tn ) b ( tn) * 4 * * x ( * * Px, Py ) 0 x ( Px 2, Py 2 ) 0 w w w * ( tn ) 0 w ( tn ) If conton (3.3.9) oes not hol, the profts of the multnatonal frm n the north wll be smaller wth FI n manufacturng than wthout t. Therefore, for FI n manufacturng to take place, a fall n the northern proft after FI n manufacturng shoul be compensate by an ncrease n profts from supplyng the southern market. Ths stuaton coul happen when seconomes of scale are hgh enough for the multnatonal frm to splt proucton between the two countres an t woul requre an atonal constrant lnkng the market sze varables. In other wors, market sze n the south shoul be suffcently hgh relatve to market sze n the north for the market seekng effect of FI n manufacturng to be stronger than the negatve factor cost fferentals effect. Consequently, conton (3.3.9) shoul hol for the postve vertcal motve of FI n manufacturng to exst. 79

88 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off The followng economc nterpretaton coul be gven to the results of Proposton 4: ) Market sze ( a, a ) : Hgher eman from the south woul apparently brng more nvestments n manufacturng whch bascally means fewer expenses per unt of the supple goo. The mpact of the market sze n the north wll epen on the presence or absence of the vertcal motve to nvest n manufacturng. We have a vertcal motve n our moel by assumng (3.3.9) an therefore the mpact s postve. 2) ubsttutablty (b): More substtutablty ncreases profts n both cases as t practcally creates bgger markets for the multnatonal. Thus, the moe wth more quanttes sol wll be chosen. 3) outhern strbuton margn ( K ): mlarly to the prevous argument but only relatve to the south, the neffcency cost s hgher uner the moe wth more quanttes sol n the south. 4) orthern wage ( w ) : A hgher northern wage means hgher cost for the multnatonal frm. Apparently, n orer to avo payng a hgher cost at home, the multnatonal wll have an ncentve to nvest n manufacturng. 5) Prouctvtes (,, ) : The prouctvty n the north s the nverse of the northern wage an the prouctvty of proucng an exporte goo n the south ncreases the ncentve to nvest. On the other han, hgher prouctvty n proucng a local goo n the south woul squeeze the market for the exporte goo an make nvestng n manufacturng less proftable. 6) outhern tarff ( t ) : o tarffs nee to be pa after the FI n manufacturng. Therefore the postve mpact represents a tarff jumpng effect. 7) outhern wage ( w ) : Even though a hgher southern wage ncreases profts uner no nvestments n the south by ncreasng ts market share, ts mpact on profts after FI n manufacturng s ambguous. Ths s because n aton to the nrect effect of gettng hgher market share there s a rect effect of payng more for the proucton of the prevously exporte goo. Hence, f the prouctvty of proucng the local goo s suffcently hgh so that t satsfes (3.3.0), the rect effect wll be stronger than the nrect effect an the total effect on profts uner the FI n manufacturng wll be negatve. mlar conseratons to the mpact of the northern tarff presente below coul explan the negatve mpact of the southern wage on the proft n the north. 8) orthern tarff ( t ) : The mpact of the northern tarff s restrcte only to the north. Even though both profts wth the FI n manufacturng an wthout wll fall uner the hgher tarff, for the gven wage rate the cost of proucton wll be etermne by prouctvtes. Hgher prouctvty means smaller cost or smaller proucer prce for the retaler. In other wors, f the prouctvty of proucng the local goo s hgh enough an satsfes (3.3.0), the proucer prce of that goo s small enough relatve to the proucer prce of the exporte goo. That makes the mpact of the tarff hke stronger on the exporte goo an therefore the profts wth the FI n manufacturng woul contract more than the profts wth no FI n manufacturng. 80

89 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off 9) Transportaton cost ( ) : The mpact of the transportaton cost s ambguous. In contrast to the northern tarff scusse above, transportaton cost oes not epen on the proucer prce an apples per unt of output. On the one han, nvestment n manufacturng reuces the transportaton cost of supplyng the south. On the other han, the transportaton cost of supplyng the north ncreases. The total effect wll epen on the relatve quanttes supple by the multnatonal to the north an the south whch, n turn, epens on the values of the market sze n the two countres. xv In other wors, a large market sze n the north relatve to the south woul lea to a negatve mpact on FI n manufacturng an a large market sze n the south relatve to the north woul lea to a postve mpact. Ether case coul happen n practce umercal results electng parameter values The analytcal results of the parameter mpact have been obtane so far by usng the constrants of the feasblty set an some ntutve assumptons but wthout assgnng any partcular values to the parameters. However, t s possble to obtan more nsghts nto our analyss by assgnng values to the parameters wthn the range mple by the emprcal evence an by choosng unts of measurement. In partcular, we wll be able to see the mpact of the FI n strbuton on the volume of trae, exports an mports. Furthermore, numercal analyss wll enable us to stuy the mpact of the southern tarff on the ncentve to nvest n strbuton. Ths effect s foun to be ambguous an wll be ealt wth n the next subsecton. Let us frst choose the reference equlbrum to be moe 4 n the south when both types of nvestments take place. The two egrees of freeom corresponng to the choce of unts of measurement of the two commotes are use n such a way that: P an P * 4 (3.4.) * x4 y As a consequence of (3.4.) we wll have. Then, we coul efne the wage per unt of prouctvty n the south as follows: w w w (3.4.2) w nce the northern wage per unt of prouctvty always appears n the form of, wthout w assgnng any value to the northern prouctvty parameter, we assume that w w. Prouctvtes are chosen to efne unts of measurement. espte hgher prouctvty n the north, the fference n the nomnal north south wage shoul be stronger. Moreover, n the prevous secton, for the ervaton of the analytcal results t was assume that: w w ( tn ) (3.3.9). Consequently, we assume that w 2 n the benchmark. Wthout a loss of generalty, we coul express all of the varables as multples of the wage per unt of prouctvty. Thus, the parameters are assgne the followng values ntally: xv The expresson for the mpact of the transportaton cost s gven at the en of Appenx. 8

90 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off w w w w, w w, a a a ( b) w, K kw, w t 0.5, t 0.05, b 0.2, 0.95 (3.4.3) From (3.4.) an (3.4.3) t follows that: * * a w Px 4 Py 4 w a (3.4.4) 2 b 2 ubsequently, w s chosen so that the prces n the reference equlbrum are equal to one: 2 w (3.4.5) a Thus, one egree of freeom s set by havng an another by etermnng the value of w from (3.4.5). xx The observables are elastctes, markups an tarffs. In partcular, the substtutablty parameter b s also the rato of cross prce to own prce elastctes of eman. All of those parameters as well as w, a, an k have no unts of measurement an shoul be assgne some emprcally observable values that coul also be checke for senstvty. Anerson an van Wncoop (2004) estmate a valorem tax equvalents of strbuton margns, transportaton costs an tarffs for a representatve nustralze country. The tarffs are estmate to be, on average, no more than 5% for evelope countres an n the range of 0-20% for evelopng countres. Ths justfes the base tarff values n (3.4.3). On the other han, transportaton costs nclue both the rect trae weghte average transportaton cost an tme cost. Anerson an van Wncoop (2004) estmate a valorem tax equvalents of those costs to be 0.7% an 9% respectvely. It as up to a 2% total transportaton cost xx. However, n our moel settng transportaton cost s gven per unt of output an not as an a valorem tax equvalent of the proucer prce. Therefore, f we are to use the value of 2% of the proucer prce as efne n Anerson an van Wncoop (2004), we wll nee to reefne the value of transportaton cost n each case. In partcular, n the south uner moe transportaton cost shoul be: w 0.2w w. mlarly, uner moe 3 t shoul be as follows: 3 3w 0.2( t ) w w. In other wors, such transformatons woul allow us to represent the consumer prce n the followng way: * r Px 3 ( )( 0.2)( t ) w w (3.4.6) Fnally, n the north transportaton cost wll look as follows: w 0.2( t ) w (3.4.7) xx If, n aton, we specfy the quantty of one goo n moe 4 to be equal to one, ue to the symmetrc eman system, the output of the secon goo wll also be equal to one. However, n that case, the value of w wll have to be negatve or the value of the substtutablty parameter wll have to be greater than one rrespectve of the values of the other parameters. Ths happens because the remanng egree of freeom was use up when we erve the eman system wth two nstea of three parameters. xx The costs are ae n a valorem terms: 0.2=(+0.07)*(+0.09)-. 82

91 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off mlarly to our moel settng, the strbuton margn n Anerson an van Wncoop (2004) consttutes both wholesale an retal trae sectors. On average, t s estmate to be equal to 55% of the proucer s prce. It s also observe that the strbuton margn coul vary more sgnfcantly for fferent goos than countres. In partcular for the U, t coul vary from 4% to 26% as of year 992. In our moel, t coul be calculate by nsertng the numercal values from (3.4.3) nto prces uner the FI n strbuton obtane n the prevous secton: It coul be shown that the value of a wll nrectly efne the value of strbuton or monopoly markup ( ). In partcular, for our reference equlbrum an for a 5 : * * * * a r Px 4 a Py 3 Py 4 Px 4 w 200% (3.4.8) 2 b w 2 Here r s the strbuton markup n the reference equlbrum an also for goo Y n moe 3. In ths case we coul also calculate the value of the strbuton markup as a share of the reference prce. For a 5, we coul fn the value of w to be: w. Therefore, the value of 3 the strbuton markup s 66.67% of the reference prce. Ths share coul be calculate smlarly n all the other cases. The markup for goo X uner moe 3 n the south s gven as follows: * * a x3 P 3 x a Px 3 w w( t ) 3w (3.4.9) 2 b w w( t ) 3 2 w ( t ) 3 39,83% 5 an w 2. x3 f a In the orth: * * a Py Pxj ( n ), 2 w t w b, 2,3, 4 j 2, 4 (3.4.0) a a a 2 r *, %, Px 2 w,,3 75% 2 t 2 b 4 In moes 2 an 4 n the south when FI n strbuton oes not take place, a wll efne a monopoly markup because the multnatonal frm stll remans the only suppler of goo X to the southern market: M M P P P * x K a ( b) K bw ( w )( t ) w M 2 ( w )( t ) w ( b)( a k) b 2 ( w )( t ) M ( b) P P * 2 x2 K a ( b) K bw w M ( a k ) (3.4.2) % M (3.4.) f a 5 an w 2 M 8, 27% M Here P s the monopoly prce charge by the multnatonal frm to local strbutors. The markups of the local goo uner moes an 2 wll be equal to k: * * Py Py 2 w( k ) y k. 83

92 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Consequently, n our numercal settng, the strbuton margn s 200% of the proucer s prce for the Y-goo an for the X-goo uner reference equlbrum. For the X-goo uner moe 3 t wll be 40% of the proucer s prce. Thus, f we assume that k 0.5, after FI n strbuton, the strbuton margn of the X-goo ecreases from the ntal value of 50%. The omestc strbuton margn an the transportaton cost coul also be calculate as a share of the reference prce: K 6.67% an 4%, 3 6,%, 7,35%. Both the strbuton margn an transportaton cost shoul be expecte to have hgher values for the representatve evelopng country than for the representatve evelope country. Moreover, transportaton costs wll epen on the estnaton country, the area of the country an the stance to ports. Thus, the benchmark values are taken to be as follows: w w w w, 2 w, a a 5( b) w, K 0.5 w, w, t 0.5, t 0.05, 3 b 0.2, k 0.5, 0.95, 0.42 w, for moe w, for moe 3 (3.4.3) w, n the orth Uner the base values of the parameters, the strbuton markup of the locally source goo Y ncreases after FI n strbuton whch correspons to scenaro (II) of Proposton 3. On the other han, hgher levels of neffcency n the southern strbuton ( K ) woul make the rse n the southern market concentraton smaller. That woul postvely nfluence the supples of both goos to the south by the multnatonal an make the fall n the prce of locally source goo X more lkely. At some pont, the case at whch both prces fall after FI n strbuton because a frm fns t more proftable to charge lower prces than before coul occur. It s more trval an correspons to scenaro (I) of Proposton 3. Furthermore, there s another pont that justfes the choce of the base parameter values to follow scenaro (II) of Proposton 3. Eckel (2009) constructs a general equlbrum moel of nternatonal retalng an fns that consumer prces wll ncrease after trae lberalzaton f the manufacturng nustry s relatvely more compettve than retalng. Our moel conforms to that fnng because manufacturng s assume to be perfectly compettve an strbutors have monopoly n sellng both goos. As t woul be seen from further analyss, trae lberalzaton brngs more FI n strbuton. Moreover, we expect the prce of the local goo to ncrease after a monopoly takes over the strbuton of t. Even though the quantty of the goo supple from the home country of the multnatonal ncreases after FI n strbuton, the mpact on ts prce s ambguous. Therefore t s not possble to show analytcally that the change n the exports from home to the host country of the multnatonal wll be unambguously postve. On the other han, a postve relatonshp between exports from the home to the host country of the multnatonal frm an FI n strbuton s generally supporte by emprcal evence as n Hea an Res (200), Belerbos an leuwaegen (998). 84

93 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Table 3.. The rato of exports from the home to host country of the multnatonal wth an wthout FI n strbuton for fferent values of K, a an b k 0. k 0.5 k 0.9 k.3 k.7 k 2. b 0. b 0.2 P x P x,70,43,24,09 0,97 0,88,09,43 * * x3 3 * * x P x P x,25,8,2,06,0 0,96,06,8 * * x4 4 * * x2 2 P x P x * * x3 3 * * x P x P x * * x4 4 * * x2 2 b 0.3 b 0.4 a 4 a 5 a 6 a 7 a 8 a 9 3,29 /A,38,43,46,47,48,49,38 /A,5,8,2,22,23,24 Here K kw an a a ( b) w. Table 3. emonstrates the rato of exports from the home to host country of the multnatonal wth an wthout FI n strbuton for fferent parameter values. If the corresponng rato s hgher/lower than one, the exports ncrease/ecrease. For most of the parameter values nclung the base values n bol, the exports o ncrease. The strbuton margn n the south leas to a fall or smaller ncrease n the prce of goo X after FI n strbuton an therefore the rato eclnes. Only at the relatvely hgh levels of southern strbuton margn, a fall n the prce s stronger than the ensung quantty ncrease. On the other han, hgher substtutablty also presente n Table 3., wll restrct monopoly power an ecrease the prce of the X-goo uner moes wth no nvestment n strbuton. Ths woul push the prces of the X-goo wth an wthout FI n strbuton closer an therefore ncrease the rato. A hgher value of market sze woul ncrease the ncentve to nvest n strbuton that results n greater value of exports to the south. A relatvely hgh substtutablty parameter (b=0,4) coul brng us out of the feasblty set as coul be seen from the corresponng column of Table 3.. Other ncators that coul be stue numercally are the value of total trae an mports from the host to the home country of the multnatonal. They woul consst of fferent components wth an wthout FI n manufacturng. In partcular, for moes an 3 the total value of trae p p wll be: P x y P an the mports from the host to the home country of the multnatonal wll be: P x y p y y. For moes 2 an 4 the corresponng terms wll be: an P p y P p x y x P x P y P x p p p x y x. Accorng to the emprcal evence n oras et al. (2007) both ncators ncrease even f the strbuton margn rses after the lberalzaton of trae n strbuton servces. However, terms relate to the prces an quanttes n the north woul be the same across both specfcatons rrespectve of the fxe cost of FI n manufacturng. Ths coul be change by mposng atonal restrctons on the resources use n the north an n the south. The problems assocate wth mposng such restrctons wll be consere n the next secton. 85

94 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off After all parameter values are assgne from (3.4.3), the expressons for all prces an quanttes woul be multples of w an all of ther multples such as values of exports, profts 2 as well as 0, 2, M, wthout fxe costs woul be multples of w. Hence, w has no mpact on the ncentves to nvest n strbuton or manufacturng. Fgure 3.. Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an fxe costs of FI n strbuton (F) For the gven values of the parameters n the feasble set, the multnatonal frm wll choose the moe that maxmzes ts profts. That choce coul be well llustrate by usng fxe costs. Consequently, Fgure 3. epcts regons for fferent values of fxe costs n manufacturng an strbuton servces an for w. The shae area of the graph represents the feasble 3 set of the values of fxe costs. Fxe costs of FI n manufacturng (FM) an strbuton servces (F) coul be thought to be partally etermne by changes n polces to stmulate or prohbt nvestments by the government n the south. Each regon n the shae area s referre to by the number whch matches the number of the moe llustrate above. In other wors, t coul be efne, for example, for moe as follows: F FM F FM F FM F FM F FM 2 3 {, (, ) (, ), (, ) (, ), 4 ( F, FM ) ( F, FM )} (3.4.4) The regon contans all the values of the parameters n the feasble set at whch the multnatonal frm receves the hghest profts uner moe. The regons corresponng to the other moes are efne smlarly. Clearly, t follows that The composton of the fgure s ntutve. The FI n both sectors wll take place uner relatvely low fxe costs. Uner a prohbtvely hgh fxe cost of manufacturng/strbuton, there s a threshol value of the fxe cost of strbuton/manufacturng such that n the area hgher than the threshol, no more FI can take place. Movement away from the north-western corner an towars the south-eastern corner of the graph woul represent a polcy shft to 86

95 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off stmulate FI n manufacturng an scourage FI n strbuton aopte by a evelopng country. The area between regon 4 an regon 3 s slghtly upwar slopng an the area between regon 4 an regon 2 s steeply upwar slopng (not vertcal) ue to the complementarty between two types of nvestments. The values of the parameters from the feasble set are such that after FI n strbuton: ) The exports from the home to the host country of the multnatonal frm ncrease n lne wth exstng evence 2) In accorance wth the general tren, the market concentraton n the south ncreases. 3) There s a vertcal motve for FI n manufacturng. We have alreay scusse what happens f one of those assumptons s lfte. In partcular, f the thr assumpton s volate an conton (3.3.9) oes not hol, then Fgure 3. woul nclue only moes an 3. The northern country tarff t s practcally equvalent to a hgher barrer for FI n manufacturng because conton (3.3.0) s assume to hol. On the other han, a hgher southern country tarff t woul rase FI n manufacturng by the so calle tarff-jumpng conseratons. For the next fgure let us assume that the tarffs for the trae n goos are entcal n both countres: t t t. Consequently, n Fgure 3.2, the regons of the feasble set for the fferent values of t an FM are llustrate. In such a case, the overall nvestment pattern wll, among other thngs, epen on the level of fxe cost n strbuton servces. If that fxe cost s hgh, only two moes of supplyng the foregn market wll be feasble. Fgure 3.2. (F = 0.6) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an blateral tarff rate (t) Apparently, at a very hgh level of a blateral tarff rate, one of the contons on the parameters wll not hol. The lne between the regons efne by an 2 s nverte U shape. As the tarff rates start fallng from relatvely hgh levels, the area corresponng to moe 2 s 87

96 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off expanng. Even though lower t makes exports to the south less expensve, the lower value of t reuces the cost of all supples back to the north uner FI n manufacturng. Whle those supples are ncreasng, moe 2 wll be expanng relatve to moe as the tarffs fall. However, at some pont, the optmal values of the supples to the north wll be acheve an the frm wll not rase ts supples any further. From that pont on, only the tarff-jumpng effect of t s prevalng an the regon of wll be expanng relatve to 2 at the lower levels of the blateral tarff ratet. Apparently, the lne between the two regons wll be non-ecreasng an concave when we use only t on the horzontal axs. Usng t for the specfcaton wth relatvely low fxe cost of strbuton (F=300) we fn a smlar expecte pattern between moes 3 an 4 n Fgure 3.3. Fgure 3.3. (F = 0.4) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n manufacturng (FM) an southern tarff rate (t ) For some values of the tarff rate n the south, there s a regon where lowerng the fxe cost of manufacturng brngs more nvestment n manufacturng an strbuton. It s, however, not ncatve of any type of complementarty between the two. Here t happens only because at the gven level of the fxe cost of FI n strbuton, moe 4 s more preferable than moe 2. 88

97 Chapter 3. FI n strbuton ervces an Trae versus Investment Trae-Off Fgure 3.4.A (FM = 0.3) an Fgure 3.4.B (FM = 0.35) Investment ecsons of the multnatonal frm for fferent values of fxe costs of FI n strbuton (F) an southern tarff rate (t ) Analogously, a graph epenng on the fxe cost of strbuton (F) an southern tarff ( t ) coul be constructe (Fgure 3.4A an Fgure 3.4B).We coul see that the threshol tarffjumpng value of t s lower wth nvestments n strbuton than wthout. That exclues the nstances of substtutablty between FI n strbuton an manufacturng. Moreover, the hgher the fference s between threshol values, the wer s the range of southern tarffs at whch FI n manufacturng s complementary to FI n strbuton wth growng F A non-monotonc relatonshp between trae costs an FI n strbuton The relatonshp between the ncentve to nvest n strbuton an the tarff rate n the south s less straghtforwar. Ths s because the mpact of t on the value of the fference n profts between moes 3 an (0) cannot be gven a certan sgn as all of the other parameters n secton 3. It coul be llustrate when we contnue wth our analyss of the prevous secton an complement Fgures 3.2 an 3.3 wth the less trval case presente n Fgure 3.5 where F= We coul see that wth an ncreasng fxe cost of strbuton from the ntal level, the regons wth no FI n strbuton wll start to emerge an expan. 89