MEASURING VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE FOR TURKISH MANUFACTURING INDUSTRY OVER TIME

Transcription

1 MEASURING VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE FOR TURKISH MANUFACTURING INDUSTRY OVER TIME A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF SOCIAL SCIENCES OF MIDDLE EAST TECHNICAL UNIVERSITY BY DEMET ŞENOĞLU IN PARTIAL FULLFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN THE DEPARTMENT OF ECONOMICS SEPTEMBER 2003

2 Approval of the Graduate School of Socal Scences Prof. Dr. Sencer Ayata Drector I certfy that ths thess satsfes all the requrements as a thess for the degree of Master of Scence. Prof. Dr. Erol Çakmak Head of Department Ths s to certfy that we have read ths thess and that n our opnon t s fully adequate, n scope and qualty, as a thess for the degree of Master of Scence. Prof. Dr. Güzn Erlat Supervsor Examnng Commttee Members Prof. Dr. Güzn Erlat Prof. Dr. Haluk Erlat Prof. Dr. Yusuf Zya Özcan

3 ABSTRACT MEASURING VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE FOR TURKISH MANUFACTURING INDUSTRY OVER TIME Şenoğlu, Demet M.S., Department of Economcs Supervsor: Prof. Dr. Güzn Erlat September 2003, 160 pages In tradtonal trade theores, foregn trade plays the role of fllng the gap of products not produced wthn the country. However, n the early 1960s ncreasng exchange of smlar products, ntra-ndustry trade, n the world trade have been observed by trade theorsts. After the realzaton of the fact that ntrandustry trade has become a very mportant part of world trade, more comprehensve studes on ntra-ndustry trade have been conducted. At the end of the 1970s, trade theorsts started to analyze ntra-ndustry trade between developed countres (horzontal ntra-ndustry trade) and ntra-ndustry trade between developed and developng countres (vertcal ntra-ndustry trade) separately, because ther characterstcs were dfferent. Horzontal ntra-ndustry models were characterzed by attrbute varaton between products whle vertcal ntra-ndustry models were characterzed by qualty varaton. Ths study nvestgates the ssue of measurement of horzontal and vertcal ntra-ndustry trade for Turksh manufacturng ndustry. We address the questons of whether the ntra-ndustry trade n Turksh manufacturng sector s more of the horzontal or the vertcal type and whether the vertcal ndustres domnates horzontal ndustres n number at the 3- dgt ndustry level. Emprcal analyses shows that the majorty of ntra-ndustry trade n Turksh manufacturng sector s of the vertcal nature; Turksh manufacturng

4 sector exports lower qualty varetes n exchange for hgher qualty varetes. Also, our emprcal analyses ndcate that a large percent of 3- dgt ndustres consdered as prmarly nvolved n ntra-ndustry trade are vertcal ndustres. Keywords: Qualty Varaton, Attrbute Varaton, Grubel and Lloyd Index, Unt Value Index v

5 ÖZ TÜRK İMALAT SEKTÖRÜNDE DİKEY VE YATAY SEKTÖR İÇİ TİCARETİN ZAMAN İÇİNDE ÖLÇÜLMESİ Şenoğlu, Demet Yüksek Lsans, İktsat Bölümü Tez Yönetcs: Prof. Dr. Güzn Erlat Eylül 2003, 160 sayfa Geleneksel tcaret teorlernde, dış tcaretn rolü ülke çersnde üretlmeyen malların boşluğunu doldurmaktır. Ama, 1960lı yılların başından tbaren, blm adamları benzer ürünlern eşanlı olarak hem hraç edlp hem de thal edlmesndek artışa, dğer br deyşle sektör-ç tcarettek artışa, dkkat çekt. Sektör-ç tcaretn dünya tcaretnn öneml br bölümünü oluşturduğunun anlaşılmasından sonra, bu konuda daha kapsamlı çalışmalar yapıldı. 1970l yılların sonunda, blm adamları gelşmş ülkeler arasındak sektör-ç tcaret (yatay sektör-ç tcaret) le gelşmş ve gelşmekte olan ülkeler arasındak sektörç tcaret (dkey sektör-ç tcaret) ayrı ayrı ncelemeye başladı, çünkü bunlar farklı özellklere sahpt. Yatay sektör ç tcaretn en öneml özellğ ürünler arasındak ntelk farkı ken dkey sektör-ç tcaretn en öneml özellğ ürünler arasındak kalte farkı d. Bu çalışmanın amacı dkey ve yatay sektör-ç tcaretn ölçümünü Türk üretm sektörüne uyarlamaktır. Bu çalışmada Türk üretm sektöründek sektör ç tcaretn çeşdnn dkey m yoksa yatay mı olduğu ve 3-basamaklı endüstr düzeynde dkey endüstrlern m yoksa yatay endüstrlern m hakm olduğu sorularına cevap aradık. Hesaplamalarımız Türk üretm sektöründek sektör-ç tcaretn çoğunluğunun çeşt olarak dkey olduğunu göstermektedr. Br başka deyşle, v

6 Türk üretm sektörü benzer ürünlern düşük kaltelsn hraç ederken aynı anda yüksek kaltelsn thal etmektedr. Ayrıca, sektör-ç tcaretn yoğun olduğu 3 basamaklı endüstrler ele aldığımızda, hesaplamalarımız bu tp endüstrlern büyük br çoğunluğunun dkey endüstrler olduğuna dkkat çekmektedr. Anahtar Kelmeler: Kalte Farklılaşması, Ntelk Farklılaşması, Grubel ve Lloyd Endeks, Brm Değer Endeks v

7 To My Famly v

8 ACKNOWLEDGMENTS I would lke to express my deep grattude to my supervsor Prof. Dr. Güzn Erlat for her nvaluable gudance and support. Studyng wth her was a great honour for me. I wll never forget her for her nvaluable support throughout ths study. My specal thanks go to Prof. Dr. Haluk Erlat for hs help and support throughout the study. I have been lucky to have spent the last three years wth hm. I wll always remember hs kndness and understandng. Thank you Professor Haluk Erlat. I would lke to thank Prof. Dr. Yusuf Zya Özcan for hs partcpaton n my masters commttee and for sharng hs valuable deas wth us. I sncerely apprecate my frend Eda Ustaoğlu, wth whom I shared three years at METU. Wth her frendly conversatons, she motvated me very much. Thanks my frend. Fnally, I would lke to thank my famly who always beleved n me for ther support, patence and encouragement. v

9 I hereby declare that all nformaton n ths document has been obtaned and presented n accordance wth academc rules and ethcal conduct. I also declare that, as requred by these rules and conduct, I have fully cted and referenced all materal and results that are not orgnal to ths work. Date: Sgnature: x

10 TABLE OF CONTENTS ABSTRACT ÖZ v DEDICATION v ACKNOWLEDGEMENTS v TABLE OF CONTENTS x LIST OF TABLES x LIST OF FIGURES xv CHAPTER 1. INTRODUCTION THE THEORIES AND THE EMPIRICAL ANALYSES OF VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE The Theores of Vertcal and Horzontal Intra-Industry Trade Models of Horzontal Intra-Industry Trade Neo-Chamberlnan Models Neo-Hotellng Models Eaton and Kerzkowsk Model Models of Vertcal Intra-Industry Trade Neo-Hecksher-Ohln Models Shaked and Suttan Model x

11 2.1.3 Concludng Remarks on Theores of HIIT and VIIT The Emprcal Analyss of Vertcal and Horzontal IIT Documentary Studes Econometrc Studes Concludng Remarks on Emprcal Analyss of HIIT and VIIT MEASUREMENT OF VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE The Measures of IIT Balassa Index Grubel and Lloyd Index The Measures of Vertcal and Horzontal IIT EMPIRICAL ANALYSIS OF VERTICAL AND HORIZONTAL INTRA INDUSTRY TRADE The Data Emprcal Results of the GL Indexes Emprcal Results of Unt Value Calculatons Emprcal Results of Decomposton of the IIT Index as HIIT and VIIT Index CONCLUSIONS x

12 REFERENCES APPENDIX A Tables x

13 LIST OF TABLES TABLE 1. Overall Summary of Founder and Determnants of HIIT and VIIT Models Aggregate GL ndexes, Dstrbuton of Number of 5-dgt Vertcal (LQ and HQ) Industres, Horzontal Industres, Inter Industres and No-Trade Industres, Dstrbuton of Number of 3-dgt Vertcal (LQ and HQ) Industres, Horzontal Industres, Inter Industres and No-Trade Industres, Decomposton of the GL Index nto ts Horzontal and Vertcal Parts SITC Codes, Numbers and Percentages of 3- dgt Industres wth more than 0,5 C Index Decomposton Results for the Year Decomposton Results for the Years 1992 and Decomposton Results for the Year Decomposton Results for the Year x

14 LIST OF FIGURES FIGURE 1. Equlbrum n the Krugman Model Alternatve Specfcatons of a Horzontally Dfferentated Commodty Plot of Aggregate GL Indexes, Dstrbuton of Number of 5-dgt Vertcal (LQ and HQ) Industres, Horzontal Industres, Inter-Industres and No-Trade Industres, Dstrbuton of Number of 3-dgt Vertcal (LQ and HQ) Industres, Horzontal Industres, Inter-Industres and No-Trade Industres, Dstrbuton of Percentages of Aggregate HIIT and VIIT Indexes, Dstrbuton of Number of 3- dgt Industres Exhbtng C Values Greater than 0, Percentages of Horzontal and Vertcal Industres n 3-dgt Industres wth C Values Greater than 0, xv

15 CHAPTER 1 INTRODUCTION Tradtonal trade theory as emboded ether n the Heckscher-Ohln- Samuelson 1 (H-O-S) postulate or n the Rcardan model s fundamentally a theory of nter-ndustry trade. Ths theory predcts that a country wll export the commodty whose producton requres ntensve use of that country s cheap factor and mport the commodty whose producton requres ntensve use of that country s scarce factors. In other words, trade s generated by supply sde dfferences. Accordngly, one would expect trade between countres wth dfferent factor endowments to be most prevalent, that s, trade, manly, between developed (the North) and developng (the South) countres (Nllson, 1999). However, n the early 1960s some trade theorsts 2 have notced that most of the world trade actually takes place between developed countres wth smlar ncome structure (Memş, 2001). Moreover, they have notced that much of the trade between developed countres nvolves two-way exchange of goods produced wth smlar factor endowments. Therefore, they hypotheszed that dfferent countres do not necessarly specalze n dfferent products but nstead dfferent countres may specalze n dfferent types of a gven commodty. As a result, ntra-ndustry trade (IIT) theory, the smultaneous exports and mports of the same statstcal product group, has emerged as a complement to nter-ndustry trade theory. 1 The H-O-S, model mples two countres, two commodtes and two factors of producton. These two countres have dfferent factor endowments and so trade occurs between these countres due to dfferences n factor prces. (Kenen, 1989). 2 See Verdoorn (1960), Lnder (1961), Posner (1961), Mchaely (1962) and Kojma (1964) 1

16 Broadly speakng, two classes of explanatons may be gven for the aggregate behavor of IIT over tme : demand spectrum aspect and supply spectrum aspect. Demand sde theores on IIT (Lnder 1961, Lancaster 1980, Helpman1981) places emphass on dversfcaton of consumer tastes. These theores predct that the extent of IIT wll be larger, the hgher and the more smlar the ncome levels of tradng partners are (Hellvn, 1996), because, when the ncome level ncreases so does the demand for dfferentated goods. Also, the more smlar the per capta ncome s, the more smlar wll be the demand structure and therefore the demand wll be larger for varetes produced n the partner country. As a result, demand sde theores pont out that, n dentcal economes dversty of consumer preferences lead to IIT (Lancaster 1979). Supply sde theores (Helpman 1981, Krugman 1981 and Helpman and Krugman 1985), on the other hand, focus on the ablty to produce a wde-range of dfferentated products. These theores predct that hgher average-per-capta ncome represents a hgher average captal-labor (K/L) endowment rato. On the assumpton that ndustres that are captal ntensve tend to have relatvely more producton of dfferentated products, countres wth hgher average K/L ratos wll experence a greater share of ntra ndustry specalzaton (Bergstrand, 1990). Therefore, as ponted out by supply sde theorsts, product dfferentaton s the most mportant source of IIT n dentcal economes. It can be concluded, from demand and supply sde theores that trade n dfferentated products are most lkely to take place between countres wth smlar factor endowments and wth hgh level of per capta ncome or, n other words, between developed countres More comprehensve studes on the development of IIT have been conducted after the realzaton of the fact that IIT has become a very mportant part of the world trade, n partcular among developed market economes. The models of IIT generated by Lancaster (1980), Krugman (1979; 1981), Helpman (1981), and Dxt and Norman (1980), horzontal ntra ndustry (HIIT) models, explan IIT by horzontal product dfferentaton (attrbute varaton between products of a smlar qualty) n combnaton wth ncreasng returns to scale. Accordng to these trade theorsts such trade may be expected to be most frequent among countres on the same level of economc development, wth the same relatve factor endowments of 2

17 resources and wth a hgher level of per capta ncome and n goods where products are horzontally dfferentated and fnally n markets characterzed by monopolstc competton. Although HIIT models state that IIT s most lkely to take place between developed countres (north-north IIT), over the perod the amount of IIT occurrng between developed and developng countres (north-south IIT) has started to grow rapdly. The realzaton of north-south IIT has become the startng pont for the development of a new IIT model, the vertcal ntra-ndustry (VIIT) model. Falvey (1981) was the founder of the VIIT model. Hs model s consdered as an alternatve to that of modelng IIT as a consequence of horzontal product dfferentaton and ncreasng returns to scale. In hs VIIT model, he explans the IIT n vertcally dfferentated products (qualty varaton between products) by retanng the constant returns to scale assumpton of the tradtonal H-O-S theorem. Accordng to hm VIIT s more lkely to take place between economes at dfferent levels of ncome. As a result, durng the 1980s the dea that IIT takes place between economcally smlar countres has lost ts valdty and a perod of horzontal-vertcal dstncton has started n IIT theory. Besdes the trade theorsts approachng the IIT subject from the theoretcal pont of vew, there are also researchers who approach the subject from the methodologcal and the emprcal pont of vew. IIT theory owes a profound debt to the efforts of Verdoorn (1960), Kojma (1964), Balassa (1966), Grubel and Lloyd (1971), Aquno (1978) and Greenaway and Mlner (1981; 1983) on the ssue of measurement of IIT and to the efforts of Aquno (1978), Greenaway (1983; 1984) and Balassa (1986) for provdng an extensve databank of evdence on recorded IIT and, fnally, to the efforts of Abd-el Rahman (1991) and Greenaway, Hne and Mlner (1994; 1995) for testng specfc hypothess related to IIT and for decomposng IIT nto HIIT and VIIT. The measurement of IIT, conductng case studes and emprcal studes on IIT and fnally decomposton of IIT as HIIT and VIIT s also applcable to the case of Turksh foregn trade. Turkey underwent major changes n her trade regme when an outward-orented growth polcy was adopted n 1980s; ths brought an expanson n exports as well as a change n composton of exports. (Erlat, 1998). The composton of exports has shfted from tradtonal to non- 3

18 tradtonal sectors 3 (dfferentated goods sector). Whether Turkey s trade structure s predomnantly nter-ndustry or ntra-ndustry and whether there has been a shft towards IIT has been nvestgated extensvely (Erlat and Erlat 2003a and Memş 2001). However, these studes have not consdered the horzontal-vertcal dstncton n IIT theory. Although there has not been gven enough mportance to ths dstncton, decomposton of IIT as horzontal and vertcal IIT s qute mportant for the Turksh manufacturng ndustry, because, as already ponted out by Abd-el-Rahman (1991), HIIT and VIIT are affected dfferently by some explanatory varables such as country-specfc and ndustry-specfc varables. If we dstngush horzontal ndustres from vertcal ones, we can determne more easly how to treat each ndustry; for example, we know that we should promote product nnovaton n horzontal ndustres and captal mprovement n vertcal ndustres. For ths reason, n ths thess, the emphass wll be on horzontal and vertcal IIT for Turksh manufacturng sector so that we can determne n whch sectors the domnant form of IIT s horzontal and n whch sectors t s vertcal. Therefore, n the followng chapter a survey of theoretcal and emprcal studes on models of HIIT and VIIT wll be presented. In Chapter 3, measures of IIT and ts decomposton as vertcal and horzontal wll be descrbed. Chapter 4 wll contan the emprcal results of applyng measurement of IIT and ts decomposton to the Turksh manufacturng sector for the perod Fnally n Chapter 5, conclusons of our case study wll be presented. 3 See Erlat (1998), Erlat and Erlat (2003b) for the classfcaton of tradtonal and nontradtonal sectors by usng Cumulatve Export Performance Functon. 4

19 CHAPTER 2 THE THEORIES AND THE EMPIRICAL ANALYSIS OF VERTICAL AND HORIZONTAL INTRA-INDUSTRY TRADE 2.1 The Theores of Vertcal and Horzontal Intra-Industry Trade Over the last twenty years, a set of new theores has been put forward to complement and/or substtute for the smple H-O-S model. H-O-S theory s based on the dea that dfferences n factor endowments between countres are the source of comparatve advantage; comparatve advantage s the determnant of nternatonal trade and, therefore reflects nter-ndustry trade. If nter ndustry trade s the domnant form of trade, then there wll be no recprocal trade of products related to the same ndustry. Accordngly, one would expect one-way trade between countres wth dfferent factor endowments (north-south trade) to be most prevalent n the H-O-S model (Nllson, 1999). However, the H-O-S model seemed ncapable of explanng certan sgnfcant emprcal fndngs about the world economy. The most mportant one of them was the Leontef paradox, namely the growth n two-way trade 4 between smlar economes wth near dentcal factor endowments (north-north trade) (Stewart, 1984). Ths ncapablty of the H-O-S paradgm n explanng two-way north-north trade was the startng pont of the emergence of a new trade theory, namely ntra-ndustry trade theory 5 (IIT). One of the earlest emprcal works demonstratng the mportance of IIT between developed countres wth smlar factor endowments was Grubel and Lloyd (GL) (1975). Accordng to GL (1975) the H-O-S model wth ts relance on relatve factor endowments as the bass for trade cannot provde a satsfactory 4 Recprocal trade of products belongng to the same ndustry 5 Smultaneous exports and mports of products belongng to the same ndustry 5

20 explanaton for the phenomenon of IIT. As a result they focused ther attenton on preference dversty as a source of trade. Ths focus has ts more mmedate orgns n the observaton about trade n Lnder (1961). Two other mportant theoretcal contrbutons were made soon after the publcaton of the GL work. Dxt and Stgltz (1977) have modeled stuatons where ndvduals value product varety and where the producton sde of the economy s characterzed by monopolstc rather than perfect competton. An alternatve model of product varety has been demonstrated by Lancaster (1979). In ths verson, products represent a bundle of characterstcs and every consumer has hs most preferred package of characterstcs. Ths, agan, gves rse to demand for varety at the aggregate level. Due to the exstence of nternal economes of scale, only some consumers are able to obtan ther deal products, others buy dfferentated products that only approxmate ther most preferred model. Both the Dxt-Stgltz and Lancaster approach to product dfferentaton rely on the exstence of ncreasng returns to scale and monopolstc competton ; none of them was compatble wth the H-O-S theorem (Kerzkowsk, 1984). These three developments (GL 1975, Dxt and Stgltz 1977 and Lancaster 1979) have ponts n common. Explanatons of trade based on preference dversty have been based on non-compettve assumptons and upon an assumpton of ncreasng returns to scale. However, the ntroducton of deas of preference dversty, mperfect competton and ncreasng returns to scale by these three developments to nternatonal trade theory have resulted n the problems that cannot be solved by the tradtonal trade theores. The theoretcal lterature on IIT theory s very extensve. It s possble to dstngush two knds of IIT, horzontal and vertcal IIT. Horzontal ntra-ndustry trade (HIIT) arses when there s a two-way trade n products of smlar qualty wth dfferent attrbutes (horzontally dfferentated products). The theoretcal bass for such trade was developed by Lancaster (1980), Krugman (1981), Helpman (1981; 1987) and Bergstrand (1990). In these models, IIT emerges n monopolstcally compettve markets wth ncreasng returns to scale on the supply sde and dverse consumer preferences on the demand sde (Mora, 2002). Also, these models suggest that the more smlar countres are n terms of ther 6

21 endowments, the greater the share of HIIT. Alternatve models of IIT nclude vertcal ntra-ndustry trade (VIIT). VIIT s trade n varetes of products characterzed by dfferent qualtes (vertcally dfferentated products). The theoretcal bass for ths type of IIT was developed by Falvey (1981) and Falvey and Kerzkowsk (1987). In these models, IIT wll take place n perfectly compettve markets but there are no ncreasng returns to scale n producton (Mora, 2002). VIIT models suggest that greater the dfference n the level of factor endowments between countres, the greater wll be the share of VIIT. The two sets of models are dfferent n ther predctons. Whereas HIIT s more lkely to take place between countres wth hgh and smlar per-capta ncomes (north-north IIT), VIIT s more lkely to take place between countres at dfferent levels of per capta ncomes (north-south IIT) (Hellvn, 1996).The dstncton between VIIT and HIIT s an mportant one. The vertcal models can explan IIT wthout recourse to economes of scale and hence, wthout serously vtatng the precepts of the H-O-S theorem. However, ths s not the case of horzontal models n whch nteracton between scale economes and (horzontal) product dfferentaton s an essental ngredent (Tharakan and Kerstens, 1995). The exstence of vertcal versus horzontal product dfferentaton does not mean that IIT cannot take place n homogenous goods. In the extreme case, two dentcal countres of the same sze wth the same taste patterns and access to the technology may stll engage n trade when trade takes place under duopoly condtons (Wllamson and Mlner (1991)). The model of Brander (1981) succeeds n showng that nternatonal trade can occur even n a world of one homogenous good produced at the same cost at home and abroad. One can thnk that ths type of trade s socally wasteful because t nvolves movng goods pontlessly across borders and uses up resources n dong so. But, ths s not the case. Although there s a socal waste of resources stemmng from transportaton costs that must be ncurred, there s also a socal gan from ncreased competton between duapolsts: consumers face lower prces after trade (Kerzkowsk, 1996). The socal value of such trade depends, therefore, on the net effect of these losses from transportaton cost and offsettng socal gan from ncreased competton. In the lterature, however, the models of IIT n homogenous goods are not as mportant as models of IIT n horzontally dfferentated goods (HIIT models) 7

22 or models of IIT n vertcally dfferentated goods (VIIT models). In the followng sectons HIIT and VIIT models wll be analyzed separately n detal Models of Horzontal Intra-Industry Trade The models of HIIT explctly ntroduce economes of scale and mperfect competton n the analyss unlke models of VIIT (Tharakan and Kerstens (1995)). Therefore, a farly large percent of HIIT takes place under monopolstcally compettve markets. The models of HIIT n monopolstcally compettve markets have n common explct assumptons of ncreasng returns to scale, free entry and ext and an assumpton that consumer preferences are suffcently dverse to ensure that a large number of sngle product frms can coexst n the fnal equlbrum (Greenaway, 1987). These models can be categorzed as neo-chamberlnan 6 and neo-hotellng 7. Both of these models exst under condtons of monopolstc competton. However, the treatment of consumer preferences dffers between these two models. In the former, the consumers endeavor to consume as many dfferent varetes as possble and more than one frm wll produce the same varety. In the latter, dfferent consumers have dfferent preferences for alternatve varetes of gven commodtes (Tharakan and Kerstens, 1995). Although the models of HIIT n monopolstcally compettve markets assume free entry and a small level of scale economes, n practce ths may not be the case. Entry may be restrcted and/or level of scale economes may be large relatve to total market demand. Under these crcumstances, the number of frms operatng n the market wll be qute small, n other words, market structure wll be an olgopolstc one. Eaton and Kerzkowsk (1984) was the frst to assert the exstence of HIIT under olgopoly condtons. Therefore, t can be sad that an olgopolstc market s an alternatve market structure under whch HIIT can take place. Accordngly models of HIIT can be categorzed as neo-chamberlnan, neo-hotellng and Eaton and Kerzkowsk model. These three models dffer from 6 For detal see Dxt and Stgltz (1977), Krugman (1979, 1980) and Dxt and Norman (1980) 7 For detal see Lancaster (1979, 1980) and Helpman (1981) 8

23 each other wth respect to market structure on whch they are based. Whle neo- Chamberlnan and neo-hotellng models are based on monopolstcally compettve markets, the other s based on olgopolstc markets Neo-Chamberlnan Models Neo-Chamberlnan models consder monopolstc competton and horzontally dfferentated goods on the supply sde. On the demand sde, model s based on love of varety approach. Accordng to ths approach all varetes enter the ndvdual s utlty functon n a symmetrcal fashon; that s, ndvduals gan utlty from greater varety, n other words, from beng able to consume more varetes, rather than from beng able to consume a preferable varety (Wllamson and Mlner,1991). Fundamental work on demand for varety was done by Dxt and Stgltz (1977) n the context of a closed economy organzed along lnes of the Chamberlnan model of mperfect competton. In ther model, trade s caused by economes of scale rather than by dfferences n factor endowments. The Dxt and Stgltz model was then appled to the open economy by Krugman n a seres of artcles (Krugman, 1979, 1980, 1982), as well as by Dxt and Norman (1980). Then, the essental features of the neo-chamberlnan model can be llustrated by reference to these authors. Kerzkowsk (1996), by reference to Krugman (1979), llustrates the essental features of the neo-chamberlnan model. In Krugman s (1979) model t s assumed that all consumers are alke and that ther taste patterns can be represented by the followng utlty functon: U = v(c ) v > 0, v < 0 (2.1) where v and v are the frst and second order dervatves of v wth respect to c and c denotes consumpton of the th good by the representatve consumer. The goods enterng the above utlty functon are produced by one dfferentated- 9

24 product ndustry; there s no need to brng an outsde homogenous good at ths stage. The dscusson refers ntally to a closed economy. Equaton (2.1) has the property that the level of utlty ncreases as the number of goods consumed by an ndvdual goes up, ceters parbus. To see ths pont the model represents the equaton (2.1) n a more specalzed form: U = θ c 0 < θ < 1 (2.2) The model supposes that ntally n goods were consumed, that ther prces were the same and equal to one, and that the representatve consumer had a money ncome of I. It follows that he must have consumed the same quantty of each good, c (n fact, c = I n ). Gven the ntal condtons, the level of utlty reached by the ndvdual was: U ( n) = nc θ (2.3) Later, the model assumes that the same consumer wth the same ncome and prces s offered nk goods nstead of n goods to consume. Through straghtforward substtuton, the dfference n the utlty levels assocated wth the two bundles can be shown as U ( nk) U ( n) = nc θ ( 1 θ k 1) (2.4) If varety s greater n the second stuaton, k > 1 and the expresson n (2.4) must be postve. By consumng less of every good ( I nk nstead of I n ) but more goods, the level of utlty attaned goes up even though one s ncome s the same and prces are unchanged. Ths s how varety makes you better off. As a second step, Kerzkowsk (1996) turns to the supply sde of Krugman s 1979 model. In ths model, only one factor, labor, s requred, and 10

25 producton functons are the same for all goods. They can be represented n terms of the number of labor unts, l, requred to produce x quantty of good : l x = α + β α, β > 0 (2.5) where l denotes labour nputs, α and β denote fxed and margnal costs respectvely. Gven that the coeffcent α s assumed to be greater than zero, there are economes of scale n producton. Wth ncreasng returns to scale l x decreases as x ncreases) there wll be only one producer of a partcular ( dfferentated good, and ths producer wll try to explot hs monopoly power over a segment of the market, n other words, t wll equate margnal revenue wth margnal cost. If all dfferentated goods are taken nto account there wll be as many producers as the number of products suppled to the market. Gven the utlty functon n (2.4) and the producton functon n (2.5), the proft maxmzng condton for a partcular producer, equatng margnal revenue wth margnal cost, becomes: P ( x )(1 1 e ) = β w (2.6) where e s the elastcty of the demand facng the ndvdual frm, and w stands for the wage rate. If new frms can be set up, no profts can be made n equlbrum even though every producer tres hard to acheve ths goal, as equaton (2.6) testfes. The zero proft condton s a property of Chamberln s model of monopolstc competton: π = P x ( α + β x ) w = 0 (2.7) From the zero proft condton t can be concluded that the total cost of producton of good has to be equal to total revenue: 11

26 P x = ( α + β x w (2.8) ) Equaton (2.8) can be put dfferently as prce equals average cost to determne the prce whch each representatve frm wll charge: α P = ( + β ) w (2.9) x Snce solutons for P, x and c wll take the same values for all s, the Chamberlnan zero proft condton can be smplfed by dsappearance of the subscrpt. Also, t can be further smplfed by settng w = 1.0. α P = + β (2.10) x Because of the fact that all varetes enter the consumers utlty functon symmetrcally and snce dfferentaton s costless, no two frms wll ever produce the same varety n the Chamberlnan model. Therefore, the output of any sngle frm wll be x = Lc (2.11) where L (2.12) = l In other words producton of a good x s equal to the consumpton of a representatve ndvdual ( c ) tmes the labor force (L ) snce the ndvduals are dentcal wth workers. By ths defnton of alternatvely as x equaton (2.10) can be wrtten 12

27 α P = + β (2.13) Lc Kerzkowsk (1996) presents the equlbrum condtons (2.6) and (2.13) n Fgure 1 as n Krugman (1979). In the fgure, the vertcal axs represents prce and horzontal axs represents per capta consumpton of every good. The PP schedule depcts equaton (2.6); t s upward-slopng because t s assumed that the elastcty of demand facng an ndvdual producer gets smaller as hs output expands. The ZZ curve n Fgure 1 shows equaton (2.13). The ntersecton of the two schedules determnes the equlbrum prce, P 0 and the level of per capta consumpton of every good, c 0. The level of output of every frm, x 0, can be obtaned by multplyng c 0 by L. The number of varetes n the economy s left undetermned. But, t can be easly determned under full employment condtons by the sze of the labour force ( L ) and the amount of labour requred ( l ) to produce a representatve varety. Under full employment condtons L = nl = n α + β x ) (2.14) ( The only unknown n ths equaton s n, the degree of product varety, and t can be calculated as L L n = = (2.15) l α + β x 13

28 P Z P P 0 e P c 0 Z c Fgure 1 Equlbrum n the Krugman Model The degree of product varety under equlbrum condtons can be presented as n *, where n * L L = = (2.16) l α + β x 0 0 Fnally, as one mght expect, symmetry n the neo-chamberlnan model ensures that all varetes are produced n the same quanttes and, n equlbrum, sell at the same prce. These features are vald for a sngle closed economy. To provde an explanaton of ntra-ndustry trade Krugman (1979) smply assumes the exstence of a second economy dentcal n every respect to the home economy. Assumng zero transport costs, two-way trade n dfferentated goods wll take place even f an dentcal range of varetes was produced n each country pre-trade. Ths follows because no frm has an ncentve to produce exactly the same good that s already produced by another frm, namely, a perfect substtute (Greenaway, 1987). Thus further product dfferentaton wll occur once trade opens: competton forces a producer from one of the two countres to ext the ndustry or to produce a new varety. Then, IIT n dfferentated products wll result (Wllamson and Mlner,1991). 14

29 Accordng to Wllamson and Mlner (1991), the characterstcs of the post-trade equlbrum wll be smlar to those of the pre-trade equlbrum: prces and outputs of all varetes wll be dentcal. No two countres have a comparatve advantage n any subset of products and yet there s a bass for trade: namely, ncreased varety. The total number of varetes ( n T ) avalable to all consumers wll now be ncreased to: ( L + L n H F n T = = h + l ) n f (2.17) where LH s the labour force at home country and LF s the labour force n the foregn country. And, n h s the number of varetes produced at home country and n f s the number of varetes produced at the foregn country. Thus, although the openng of trade has no effect on the level of output or the number of frms producng n ether country, consumers have twce the number of products avalable. Gven ther preference for dversty, they consume some of all varetes. If nt varetes are consumed by every one at home and at the foregn country, but only n = n are produced n each country, then there must be two-way trade h f between the two countres through the exchange of varetes. Gans from trade n ths partcular model of HIIT are the outcome of ncreased varety. Ths s a consequence of the specal form of the utlty functon adopted. If per unt costs decrease as the scale of producton ncreases, gans could come smultaneously from ncreased varety and from reduced prces of domestc and foregn-produced dfferentated goods followng the enlargement of markets, whch results from the openng of trade (Wllamson and Mlner,1991). However, several lmtatons of ths model can be dentfed. Although the posttrade number of varetes can be specfed, the locaton of frms, and therefore the pattern of trade are ndetermnate. Also, nothng can be sad about whch varetes wll be produced domestcally and whch wll be mported. Helpman and Krugman (1985) have reacted to ths shortcomng by accommodatng dfferences n ntal factor endowments. In such a case also, as long as some frms n both 15

30 economes produce dfferentated goods, IIT wll take place snce the monopolst producer wll sell n both economes Neo-Hotellng Models The neo-hotellng model, as n the neo-chamberlnan case, s also based on monopolstc competton and horzontally dfferentated products on the supply sde of the economy. However, from the demand spectrum, consumer preferences characterstcs of the neo-hotellng model are completely dfferent from those of the neo-chamberlnan model. Whle the Dxt-Stgltz-Krugman model s based on love of varety approach, the neo-hotellng model questons ths approach and ntroduces a new one; the deal varety approach. Accordng to the deal varety approach, ndvduals have dfferent most preferred locatons, n other words mxes of attrbutes, and each ndvdual consumes only hs or her most preferred varety, or the nearest avalable. Under ths approach ndvduals gan utlty from beng able to consume preferable varety. The alternatve, Chamberlnan, love of varety approach s to clam that all varetes enter the ndvdual s utlty functon n a symmetrcal fashon. Indvduals gan utlty from consumng more varetes, rather than from beng able to consume a preferable varety as n the case of deal varety approach (Wllamson and Mlner, 1991). Fundamental characterstcs of the neo-hotellng model have been constructed by Lancaster (1980) and Helpman (1981). In the basc neo-hotellng model, neo-hotellng model under closed economy condtons, consumer preferences are unformly dstrbuted around a crcle. Decreasng costs n producton ensures that the number of varetes produced s fnte and ths n turn ensures that some consumers obtan ther deal varety, others consume a varety that s not deal, or do not consume at all. When a consumer becomes further far away from hs deal varety, the prce he wll be wllng to pay falls n proporton to hs dstance from the deal varety. From the producton pont of vew, closed economy equlbrum condtons of the neo-hotellng model are very smlar to those of the neo- Chamberlnan model. The proft maxmzng soluton for any ndvdual frm s acheved when margnal revenue s equal to margnal cost (zero proft condton). The proft maxmzng behavour, free entry and decreasng costs determne the number of varetes produced ( n ) n the economy. All of these varetes have an equal market share and, therefore, sell at the same prce. Ths equlbrum wth n frms 16

31 producng n varetes and earnng zero profts s descrbed by Lancaster (1980) as perfect monopolstc competton (Greenaway, 1987). Internatonal trade condtons of the neo-hotellng model are brought nto pcture by Kerzkowsk (1996) wth reference to Lancaster (1980). Under trade condtons, two economes dentcal n all respects were consdered. Each economy has two sectors; a homogenous product ndustry and a dfferentated product ndustry. Whle the homogenous good s produced under constant returns to scale, there are economes of scale to be acheved n the dfferentated product ndustry. There are no barrers to entry and hence profts must be drven to zero n equlbrum. Under ncreasng returns to scale technology, only a lmted number of dfferentated products can be produced although an nfntely large number of them are demanded. Suppose that pre-trade condton of home country was as depcted n the Fgure 2. Each pont on the crcle represents an deal model for some ndvduals. If four models ( m 1, m2, m3, m4 ) are produced under closed economy condtons, the consumers wth deal models c 1, c2, c3, c4 were very lucky; they got exactly what they lked best. Others pay the same prce for models that dd not correspond to ther most preferred specfcatons. Suppose now that free trade takes place between the home country and n an dentcal foregn county. There are also four foregn frms but, nstead of producng, m 1, m2, m3, m4, they happen to produce four other models, one of whch s halfway between m 1 and m2, one between m 2 and m3, and so on. It can be seen that free trade would be benefcal for some consumers and harmful to none, because, some consumers wll be closer to ther deal model as the number of frms, each producng a dfferent varety, ncreases. Trade wll be entrely of ntrandustry character n such a stuaton. 17

32 m, c 1 1 m, c m, c m 3, c 3 Fgure 2 Alternatve Specfcatons of a Horzontally Dfferentated Commodty Accordng to Kerzkowsk (1996) when two countres are lke twns, trade wll nevertheless take place and t wll be entrely IIT. However, n case of dentcal economes nothng can be sad about whch varetes wll be produced domestcally and whch wll be mported. Ths problem was also the most mportant shortcomng of the neo-chamberlnan model. The neo-hotellng model has reacted to ths shortcomng by accommodatng dfferences n ntal factor endowments and asserted that factor endowments may dffer between countres and the dfferentated product ndustry may be relatvely captal ntensve whle the homogenous product ndustry s relatvely labour ntensve. In such a case, as Lancaster (1980) and Helpman (1981) have already shown, ntra- and nterndustry trade wll coexst. Both countres wll be exportng and mportng smultaneously the dfferentated product; however, the country wth the hgher overall captal-labour rato wll be a net exporter of the dfferentated product whle the other country s a net mporter. In order to balance the trade, the relatvely captal poor country wll be the only exporter of the homogenous product. Therefore, there wll be un-drectonal trade n the homogenous good sector whle there s a b-drectonal trade n the dfferentated product sector snce each good wll be produced n only one country. As a result, IIT n a captal ntensve, dfferentated product determned by economes of scale and monopolstc competton, coexsts wth nter-ndustry trade n a labour ntensve, homogenous product drven by cross country dfferences n relatve factor endowments (Nlsson 1999). Other thngs beng 18

33 equal, the greater the dfference between ntal factor endowments the smaller the share of IIT n overall trade Eaton and Kerzkowsk Model Eaton and Kerzkowsk (1984) have developed an IIT model that s based on olgopolstc markets and horzontally dfferentated products. The model has been the proof of the dea that an olgopolstc market s an alternatve market structure under whch HIIT can take place. Before brngng nto the pcture nternatonal trade under olgopoly, Eaton and Kerzkowsk (1984) have constructed the basc features of the model under autarky. They have used Lancaster s (1971) formulaton to characterze demand for the horzontally dfferentated products. In ths formulaton each consumer has an deal varety of good B (dfferentated product), whch can be characterzed by a parameter θ. An ndvdual wll purchase an alternatve varety that devates from the deal f the prce of the alternatve s suffcently lower. Then the utlty functon wll be { Y - p Z Y p} V ( Y, p, θ, Z ) = max θ, (2.18) where Z stands for the varety consumed by an ndvdual ; p s the prce of the dfferentated product; and Y denotes ndvdual s ncome. The utlty functon (2.18) has the followng characterstcs: At most one unt of dfferentated product wll be bought. The maxmum prce that an ndvdual s wllng to pay for t s p provded that the avalable varety corresponds exactly to θ, ths prce falls lnearly wth the dstance θ Z. When the prce for all avalable dfferentated products exceeds p θ Z the consumer wll spend hs entre ncome on good A (homogenous product). From the producton pont of vew, the producton of good B s characterzed by ncreasng returns to scale. The total cost of producng x unt of the dfferentated good B s 19

34 C = k + cx (2.19) where c s margnal cost and k represents fxed cost. In contrast wth the lterature n monopolstcally compettve markets, a frm ncurs the fxed costs when t chooses a varety to produce, before t decdes on the level of output and prce. Entry and prce decsons are thus taken sequentally rather than smultaneously. An essental pre-requste to generatng an olgopoly soluton s the restrcton that Eaton and Kerzkowsk (1984) mpose on the number of varetes that are demanded. Two cases are consdered, frst where there s only one type of consumer that demands a specfc varety; second, where there are two types of consumers each demandng a dfferent deal varety. In the two-consumer case, there wll be no more than two frms, each specalsng n one varety, n equlbrum. In order to reach ths concluson, rules concernng entry have been specfed. Entry nto the market for dfferentated good s unrestrcted. If an entrant beleves that t wll make proft t can enter. However, as already mentoned, entry s sequental rather than smultaneous. One frm decdes to enter the market wth a specfc varety; the subsequent frm take ths varety as gven and decdes whether or not to enter. Takng the two-consumer case, n 1 consumers demand varety θ 1 and n 2 consumers demand varety θ 2. The varetes θ 1 and θ 2 are consdered to be horzontally dfferentated. The number of frms n the market depends on k, c, n, n 1 2, prces of dfferent varetes, and the economc dstance or n other words, the extent of the dfference of the producton functons between θ 1 andθ 2. If k and c are very large but n 1, n2 and p are very low, then a new entry would be unproftable. However, f k and c are suffcently small and n 1, n2 and p are suffcently large, then a new entry would be encouraged. As long as two ncumbent frms are producng varetes θ 1 andθ 2 further proftable entry s mpossble, snce Bertrand prce competton drves the prce to margnal cost of at least one frm. As a result, n equlbrum, there wll be no more than two frms n two-consumer case n the Eaton and Kerzkowsk model (Greenaway, 1987). 20 After gvng the basc features of a sngle economy under autarky, Eaton and Kerzkowsk (1984) provded an explanaton of a shft from autarky to trade by ntroducng a second economy. They realsed that when trade opens up, the

35 number of varetes traded, the number of frms producng those varetes and the magntude of the net benefts of trade clearly depend on the ntal assumptons made wth regard to autarky n the tradng economes or more specfcally depend on the autarkc dstrbuton of preferences n each country. The dstance on the product spectrum between pre-trade and post-trade varetes and whether there s any overlap s also mportant. If, for nstance, economes are dentcal, then the openng of trade wll result n a sngle producer of each varety, wth each beng sold at a lower prce. In the event that one producer s located n each country we would clearly have IIT n horzontally dfferentated products, or n other words HIIT. In the Eaton-Kerzkowsk model, however, ths need not necessarly follow, because there s a homogenous good sector and, n some crcumstances, one country can specalse n both varetes of the dfferentated good, exchangng ths for the homogenous good (Greenaway, 1987). In ths case the drecton and type of nternatonal trade wll not be IIT but t wll be nter-ndustry trade Models of Vertcal Intra-Industry Trade The dstncton between vertcal and horzontal models s an mportant one. HIIT models are usually expressed as a way of explanng IIT flows between developed countres. However, IIT between developed and developng countres, namely VIIT, may be expected to be of a dfferent knd and caused by other factors than IIT among developed countres themselves. That s to say, VIIT models requre a modfcaton of the usual explanatons. An essental and nnovatve element n VIIT models s the postulaton of vertcal product dfferentaton by qualty as the crucal determnant n IIT between developed and developng countres. Furthermore, vertcal models can explan IIT wthout recourse to scale economes and mperfect competton and hence, wthout serously vtatng the precepts of the H-O-S theorem. Ths s not the case for the horzontal models n whch the nteracton between economes of scale, (horzontal) product dfferentaton and mperfect competton are essental ngredents (Tharakan and Kerstens, 1995). 21

36 It can be sad that a farly large percent of VIIT takes place n perfectly compettve markets. The frst explct paper concernng VIIT n perfectly compettve markets s Falvey (1981). In ths paper, Falvey showed that VIIT may arse n stuatons where large numbers of frms produce varetes of dfferent qualtes wthout ncreasng returns n producton. By ths way he has extended the H-O-S theorem to construct the neo-hecksher-ohln model. Although not as wdespread as the neo-hecksher-ohln model, Shaked and Suttan (1984) has constructed a VIIT model n whch the number of frms s endogenous. In ther 1984 paper Shaked and Suttan showed that VIIT may also arse n a market structure wth small numbers of frms and ncreasng returns to scale unlke the neo-hecksher-ohln model. These two models of VIIT can be dstngushed dependng on the market structure on whch they are based. The Neo-Hecksher-Ohln model exsts under perfectly compettve markets, whle the Shaked and Suttan model exsts under natural olgopoly Neo-Hecksher-Ohln Models Ths approach can be consdered as an alternatve to that of modellng IIT as a consequence of scale economes and monopolstc competton. The frst paper on ths approach s Falvey (1981). Falvey (1981) attempts to mnmze the departure from the tradtonal H-O-S theorem by modfyng the standard framework n a mnor fashon. In the tradtonal H-O-S model, two factors are used to produce two commodtes n two countres. Ths model assumes that dfferental factor endowments that cause autarkc factor prce dfferences between the potental tradng partners are the reason for trade. There s also the assumpton of constant returns to scale n the H-O-S model. Falvey (1981) retans these two central assumptons of tradtonal theory. However, for the sake of extendng the H-O-S model, he makes two crucal modfcatons. Frst, he assumes that one of the two factor nputs used n each ndustry (captal) s specfc to that ndustry. The second modfcaton s that each ndustry s no longer assumed to produce a sngle homogenous product, but nstead can produce a range of products usng as nputs labour and ts own ndustry specfc captal, n other 22

37 words, at least one ndustry s assumed to produce a dfferentated commodty. The commodty concerned s vertcally dfferentated, dfferentated wth respect to qualty (Greenaway, 1987). Falvey (1981), after touchng upon hs basc modfcatons, constructs the closed economy features of the Neo-Heckscher-Ohln model. The ndustry under consderaton s assumed to posses a gven stock of captal ( K ) and to be able to hre labour ( L ) at the gven wage rate ( w ). Usng the servces of K and L, the ndustry can produce a range of products, whch wll be referred to as dfferent qualtes. On the supply sde, product qualty s determned by the captal-labour rato ( α ) used n the producton. Hgher qualty products requre more captalntensve technques and therefore have hgher prces. On the other hand, from the demand sde, demand for each qualty s taken to be a functon of the prces of all qualtes and total consumer ncome. Consumers are assumed to prefer hgh qualty to low qualty products. Snce, however, choce s ncome constraned some consumers wll ntally be confned to some low qualty varety wth substtuton towards hgher qualtes resultng from ncome ncreases (Greenaway, 1987). Trade condtons of the neo Heckscher-Ohln model are agan explaned by Falvey (1981). Accordng to Falvey (1981) trade takes place n a two-country (home and foregn) world, n each of whch the ndustry under consderaton has a gven captal stock (K and K*, respectvely) and faces gven wage rates (w and w*, respectvely). Captal s ndustry specfc and nternatonally mmoble, but s freely moble n the producton of ths ndustry s varous qualtes n each country. The returns to captal (r and r*, respectvely) adjust so as to mantan the full employment of the two captal stocks. Each ndustry s assumed to be perfectly compettve. Then, for any gven returns to captal n the two countres, domestc * producton costs () c and foregn producton costs ( c ) for a gven qualty α can be represented as c = w + α r (2.20) c * * * = w + α r (2.21) 23