Latin America s trade specialization and China and India s growth. Daniel Lederman Marcelo Olarreaga Eliana Rubiano

Transcription

1 Lain Ameria s rade speializaion and China and India s growh Daniel Lederman Marelo Olarreaga Eliana Rubiano Absra This paper explores he exen o whih he rapid growh of China and India in world markes is affeing Lain Ameria s rade speializaion paern. To his end we onsru a measure of Revealed Comparaive Advanage (RCA) by 3-digi ISIC seor, ounry and year, whih aouns for boh impors and expors (Vollrah, 1991). We hen explore he orrelaion of hese RCA measures beween LAC and he wo Asian eonomies. Resuls sugges ha he speializaion paern of LAC eonomies wih he exepion of Mexio has been moving in opposie direion o he rade speializaion paern of China and India. Labor inensive seors (boh unskilled and skilled) are he more likely o be negaively affeed by he growing presene of China and India in world markes, whereas naural resoure and sienifi knowledge inensive seors are likely o benefi from China and India s growh. Keywords: RCA JEL lassifiaion: F10, F14 We are graeful o Gordon Hanson, Guillermo Perry and Peri Silva for ommens and suggesions on an earlier version. This paper is par of a broader proje in he World Bank s Chief Eonomis Offie for Lain Ameria and he Caribbean o undersand he impa on LAC eonomies of he rapid growh of China and India. The views expressed here are hose of he auhors and do no neessarily refle hose of he insiuions wih whih hey are affiliaed. Developmen Researh Group, The World Bank, 1818 H Sree, NW, Washingon DC 20433, USA. dlederman@worldbank.org. Chief Eonomis Offie for Lain Ameria and he Caribbean, Bano Mundial, Carrera 7, no , Piso 16, Bogoa, Colombia, and CEPR, London, UK. molarreaga@worldbank.org, Chief Eonomis Offie for Lain Ameria and he Caribbean, Bano Mundial, Carrera 7, no , Piso 16, Bogoa, Colombia, and Ponifiia Universidad Javeriana, Bogoa Colombia. erubiano@javeriana.edu.o.

2 Inroduion In 1980 he Lain Ameria and Caribbean region (LAC) was wie as large as China and India, whih joinly represened 3 peren of world GDP. By 2004, due o he relaive rapid growh of he Asian eonomies, LAC was 20 peren smaller han China and India. 1 Today China is he sixh larges eonomy in he world when measured in erms of GDP and India he 10 h larges eonomy. Togeher hey aoun for 6.4 peren of world GDP, wih China being roughly hree imes larger han India. The fas eonomi growh of China and India was also aompanied by heir rapid inegraion o world markes, while Lain Ameria s inegraion o world markes lagged behind. LAC had a rade o GPD raio roughly equal o he rade o GDP raio of China in he lae 1980s, and wo imes larger han he rade o GDP raio of India. By 2004, he rade o GDP raio of China was 35 peren larger han he rade o GDP raio of LAC, and India s rade o GDP raio was only 14 peren smaller han LAC s. China is he hird larges rading eonomy in he world (jus behind he Unied Saes and Germany), while India ranks 25 h in erms of rade value. The fas growh of China and India and heir rapid inegraion ino world markes has probably affeed he rade-speializaion paerns of oher eonomies. They onsisenly have been aking marke share away from Lain Amerian exporers in world markes, bu he growing demand in China and India may benefi oher exporers. The imporane of China and India as a desinaion for Lain Amerian expors inreased four-fold sine 1990 when hey represened less han 1 peren of LAC expors. This signals a signifian inrease in opporuniies, even hough he levels remain quie low generally represening less han 10 peren of oal impors (see Figure 1). On he oher hand, he share of China and India in oal LAC impors also inreased signifianly over he period, signaling ha heir growing presene may be huring some firms in LAC (see Figure 2). 1 Noe ha by 1990 Lain Ameria was sill 65 peren larger han China and India. Sine hen Lain Ameria s GDP grew a an average annual rae of 4.4 peren, India a 5.7 peren, and China a 12.9 peren. 1

3 The objeive of his paper is o explore he exen o whih Lain Amerian eonomies are being pushed ino a differen rade speializaion paern by he growing presene of China and India in world markes. Are LAC eonomies adjusing? And if yes, are heir expors beoming more inensive in naural resoures, sienifi knowledge, skilled or unskilled workers? How does he adjusmen paern differ wihin LAC? The answers o hese quesions should help poliy markers aompany he adjusmen proess wih long and shor erm poliy insrumens, suh as eduaion, ehnial raining, innovaion poliies, and perhaps rade-adjusmen assisane programs for workers. Regarding ompeiion in hird markes, here is a small and growing lieraure ha argues ha here is no srong rade ompeiion beween China and LAC, wih he exepion of Mexio and some Cenral Amerian ounries (e.g., Blazquez-Lidoy, Rodriguez and Saniso, 2006). These auhors based heir onlusion on he low expor similariy aross indusries beween Lain Ameria and China. 2 There are wo problems wih he indies of expor similariy. Firs, hey fail o apure he imporane of eah produ in world markes. Tha is, China and LAC an have very differen shares in produs ha are heavily raded in world markes, and very similar ones in produs ha are no heavily raded, hus undersaing he degree of similariy. In oher words, by fousing on he wo ounries shares on heir own expors, hey fail o apure he relaive imporane of heir bilaeral ompeiion vis à vis heir ompeiion wih oher exporers. Seond, by fousing only on expors hey fail o apure wo imporan phenomena: he growing inra-indusry rade in inermediae goods and he opporuniy ha he growing eonomies of China and India represen for LAC eonomies. These seond aspe is disussed wihin he Argeninean onex by Casro, Tramuola and Mona (2005), who argued ha an imporan share of he reen expor boom in ommodiies experiened by Argenina is due o China s growing demand for ommodiies over he las 15 years (see Figure 3). Evidene of he imporan impa ha he growh of China and India had on ommodiy markes is also provided by Calderon (2006). Using a graviy ype model, Lederman, Olarreaga, and Soloaga (2006) also provide esimaes ha sugges ha he 2 More preisely heir index of expor similariy is given by s n s, whih wih he exepion of Mexio are all below 0.5. n LAC China n 2

4 growh of China, in pariular, and India o a lesser exen, had a signifian posiive impa on LAC expors. 3 In order o address he wo problems assoiaed wih expor similariy indies, and provide some seor level evidene of how China and India s growing presene in world markes may be affeing he speializaion paern of LAC eonomies we proeed as follows. Firs, we onsru an index of revealed omparaive advanage (RCA) a he global level ha aouns for expors, bu also impors, as well as he relaive size of world markes o apure he overall ompeiiveness of eah ounry by seor (Vollrah, 1991). 4 We hen explore he orrelaion beween LAC s RCA on he one hand, and Chinese and Indian RCAs on he oher hand. These exerises provide an idea of he exen o whih LAC is ompeing in he same markes as China and India, as well as wheher Chinese and Indian markes represen opporuniies for LAC expors. We also explore he evoluion of hese indies over ime o analyze wheher ompeiion and/or opporuniies have been inreasing over he las wo deades. Finally, and afer deermining he faor inensiies of LAC s eonomies in eah seor (see Appendix), we sysemaially explore he direion in whih Chinese and Indian growh is shaping he speializaion of LAC in erms of use of faors of produion. Our resuls sugges ha he speializaion paern of LAC eonomies wih he exepion of Mexio has been moving in opposie direion from he rade speializaion paern of China and India. This indiaes ha LAC s rade paerns are beoming inreasingly omplemenary o he speializaion paern of China and India. China s speializaion paern a he end of he period sudied (early 2000s) was negaively orrelaed wih he speializaion paern of mos LAC eonomies (again Mexio being an exepion). India s speializaion paern, however, seems o be posiively orrelaed even a he end 3 There are wo oher problems wih indies of expor similariy. Firs, similar produs an be expored o differen markes, hus represening lile ompeiion in speifi markes. Seond, expors of similar produs o he same marke an be subje o demand omplemenariies, due for example o he growing presene of produion neworks. Evidene provided by Lederman, Olarreaga, and Soloaga (2006) sugges ha hese omplemenariies an be large, a leas a he aggregae level. However, hese wo problems will no be addressed in his paper. 4 In order o be able o ompare RCAs aross ime and ounries we do a simple orreion o Vollrah s index, whih is disussed in he nex seion. 3

5 of he period wih hose of mos LAC eonomies (wih he exepion of he Andean ounries). In erms of he opporuniy ha he inernal marke of hese wo eonomies may represen for LAC eonomies, we found no evidene ha heir bilaeral rade has a signifian impa on LAC s speializaion paern, beyond he impa ha China and India s rade paern wih he world has on LAC s speializaion paern. This is probably due o he sill relaively small size of he bilaeral rade beween LAC eonomies, and China and India. Finally, we found ha labor (boh unskilled and skilled) was he faor ha was more likely o be negaively affeed, whereas naural resoures and sienifi knowledge were likely o benefi from he growing presene of China and India in world markes. The res of his paper is organized as follows. Seion 2 provides prima faie evidene regarding he level and evoluion of he orrelaion beween he Lain Ameria s RCAs and he RCAs of he wo Asian eonomies. Seion 3 desribes he more sysemai empirial mehodology we used o idenify he direion China and India are pushing he speializaion paern of LAC eonomies, and disusses he resuls. Seion 4 onludes. 2. How similar are LAC, Chinese, and Indian paerns of rade speializaion? Prima faie evidene In order o answer he quesion above, we sar by onsruing an index of Revealed Comparaive Advanage for China, India and hireen LAC eonomies for whih here is reliable rade daa available a he seor level (ISIC 3 digi) from 1990 o We hen explore he evoluion of China, India and Lain Ameria s eonomies RCAs over he las 15 years o provide prima-faie evidene regarding he exen o whih China and India s speializaion paern is orrelaed wih he speializaion paern of LAC. 2.1 Consruing RCA indies 5 These hireen eonomies are Argenina, Bolivia, Brazil, Chile, Colombia, Euador, Cosa Ria, Guaemala, Mexio, Niaragua, Peru, Uruguay and Venezuela. Togeher hey represen 98 peren of he Lain Ameria s GDP. 4

6 The measure of RCA we used is he one advaned by Vollrah (1991), whih orres for hree problems assoiaed wih he radiional Balassa s measure of RCA. 6 Firs, i eliminaes any double ouning problem by exluding he seor, and ounry rade values in he aggregaes ha are used as benhmarks o ompare a ounry/seor RCA. Seond, i is based on a measure of ne expors, whih allows he RCA o apure he growing imporane of inra-indusry rade. Third, Balassa s index is asymmeri as i varies beween 0 and infiniy, wih values beween 0 and 1 indiaing ha he ounry does no have a omparaive advanage and values beween 1 and +infiniy signaling ha he ounry has a omparaive advanage in ha seor. The measure proposed by Vollrah (1991) is symmeri wih posiive values indiaing a revealed omparaive advanage and negaive values a revealed omparaive disadvanage. The symmery of he RCA index is an imporan feaure for our eonomeri esimaes in he nex seion. More formally, he RCA proposed by Vollrah (1991) is given by: s ( RXA ) ( RMA ) RCA, = ln ln, (1) s, s where s ( X s, ) ( X s, ) ( X s, ) ( X s ) ( M ) ( M ) ( M ) ( X ) RXA, =, (2) RMA, (3) s = s, s, s, s, X s, X s, where are expors of ounry in seor s a ime, are oal expors of ounry X s, minus expors of good s a ime, is world expors in seor s a ime, minus, X s, X s, X s, and is oal world expors minus and. M sands for impors and subsrips and supersrips are defined in he same way as in he ase of expors. X s, 6 Balassa s measure of RCA of ounry in seor s is given by: w w ( X X ) ( X X ) RCA = s s, where are expors of ounry in seor s, X are oal expors of ounry, is world rade in seor s, and w X is oal world rade. s w X s 5

7 There is one problem wih Vollrah (1991) index if one needs o make omparisons aross ounries and ime. The average value of RCA, s aross seors s will vary aross ounries and ime. The average value will depend on he degree of onenraion of expors and impors in eah ounry/year. So in order o make inferenes regarding whih ounry has a sronger omparaive advanage in apparel, or wheher a ounry s omparaive advanage in apparel has inreased hrough ime, we need o normalize all RCA, s values by heir ounry/year mean. More formally, he measure of RCA we will be using is given by: RCA s RCA ˆ, s, = RCAs, n (4) s where n is he number of seors s. Table 1 provides summary saisis for R CA ˆ s, alulaed aording o (4) by ISIC 2 digi seor in he years 1990 and 2004 for LAC as a whole, he Andean ounries (Bolivia, Colombia, Euador, Peru and Venezuela), Cenral Ameria (Cosa Ria, Guaemala and Niaragua), Mexio, he Souhern Cone (Argenina, Brazil, Chile and Uruguay), China, and India. A quik look a Table 1 suggess ha in 1990 LAC s omparaive advanage (when RCA ˆ s, > 0 ) laid in ommodiies: agriulure (ISIC 11), logging (ISIC 12), fishing (ISIC 13), rude peroleum (ISIC 22), ore mining (ISIC 23), food manufauring (ISIC 31), basi meal indusries (ISIC 37) and eleriiy and gas (ISIC 41). 7 In 1990 China s omparaive advanage was relaively similar. I was onenraed in logging (ISIC 12), fishing (ISIC 13), oal mining (ISIC 21), rude peroleum (ISIC 22), oher mining (ISIC 7 Their omparaive disadvanage laid on some ommodiies suh as oal mining (ISIC 21) and oher mining (ISIC 29), bu mainly in manufauring: exile and apparel (ISIC 32), wood produs (ISIC 33), paper produs (ISIC 34), hemials (ISIC 35), non-meal produs (ISIC 36), fabriaed meal produs (ISIC 38), and oher manufauring (ISIC 39). 6

8 29), food manufauring (ISIC 31), exile and apparel (ISIC 32), non-meal produs (ISIC 36), and oher manufauring (ISIC 39). Thus of he eigh wo digi seors on whih LAC had a omparaive advanage in 1990, here were four on whih China also had a omparaive advanage. The omparaive advanage of India in 1990 was also relaively similar o he one observed in China, and o a lesser exen in LAC. I was onenraed in agriulure (ISIC 11), logging (ISIC 12), fishing (ISIC 13), ore mining (ISIC 23), food manufauring (ISIC 31), exiles and apparel (ISIC 32), non-meal produs (ISIC 36) and oher manufauring (ISIC 39). Thus, of he eigh seors on whih India had a omparaive advanage in 1990, here were six on whih China also had a omparaive advanage and five on whih LAC also had a omparaive advanage. However, beween 1990 and 2004 he speializaion paern of LAC and he wo Asian eonomies seemed o have moved in opposie direions. By 2004 China and LAC only shared wo seors on whih hey boh had a omparaive advanage. Wih India, he number of seors remain unhanged, suggesing a less pronouned deline. Bu Table 1 also provides informaion regarding indusries ha have experiened a signifian hange in omparaive advanage (of more han 0.5 poins in heir RCA) by indiaing he values of he RCAs in hose indusry/regions in bold when here is an large inrease, and by undersoring he values of he RCA in he presene of large falls. An impressionisi look a Table 1 suggess ha India and China had large inreases in heir omparaive advanage in he manufauring seor, and large falls in heir omparaive advanage in ommodiies (agriulure, fishing, logging, and mining). In LAC on he oher hand, hanges have been raher modes, suggesing a more sable speializaion paern. Bu LAC s relaively sable speializaion paern hides some ineresing regional differenes. For example, Mexio and o a lesser exen Andean ounries and Cenral Ameria have a similar paern o he one observed in China and India, alhough no as pronouned. For he Souhern Cone, he opposie is rue, as hese ounries seemed o have experiened large inreases in heir omparaive advanage for ommodiies, while he saw heir omparaive advanage in manufauring experiene some large falls. 7

9 Overall his firs look a he daa suggess ha he speializaion paern of LAC in 1990 was relaively o he one of China and India. However, on aggregae LAC s speializaion paern seemed o have been moving apar from China and India s who have shifed owards manufauring indusries, while LAC s speializaion paern remained relaively sable. However, his hides some ineresing differenes aross regions, wih Mexio moving in he same direion as China and India, while he Souhern Cone moved in opposie direion. The evoluion of Andean ounries and Cenral Ameria s speializaion paern is in beween he evoluion of he wo oher LAC regions. 2.2 Correlaion beween LAC s and Chinese and Indian RCAs To see wheher LAC is ompeing in he same produs wih China and India in a more sysemai manner, we alulae he orrelaion beween LAC s RCAs and hose of he wo Asian eonomies by year. Posiive values for he orrelaion of RCAs indiae ha ounries end o have similar speializaion paerns, and herefore will be ompeing in he same markes. Negaive values for he orrelaion of RCAs indiae ha he speializaion paerns omplemen eah oher, and ha he growh of China and India is an opporuniy for Lain Amerian firms, as China or India s ne impor demand will be large when LAC s ne expor demand is large. We also explore he evoluion of he orrelaion beween 1988 and Figure 4 shows he evoluion hrough ime of he orrelaion beween Chinese and Indian RCAs and he RCAs for he aggregae of hireen LAC ounries in our sample. The blue line shows he orrelaion wih China and he red line he orrelaion wih India (we will disuss he green line in seion 2.3 below). A he beginning of he period he orrelaion beween Chinese and Indian RCA s on he one hand, and LAC RCAs was posiive bu modes (around 0.2), suggesing ha China and India were speializing in he same produ as LAC. However, he rend is learly downwards and by he end of he period, 8 I is imporan o noe ha he growing presene of China and India an be a hrea or a omplemen for LAC expors, regardless of he signs of hese orrelaions. Indeed, here may be demand omplemenariies beween goods offered by LAC on he one hand, and China and India on he oher. Thus more expors of China and India may lead o more expors of LAC of he same produ. See Lederman, Olarreaga and Soloaga (2006) for evidene of his being he ase a he aggregae level. 8

10 he orrelaion wih China was around -0.2 and he orrelaion wih India was lose o zero. This suggess ha by he end of he period LAC s rade speializaion paern was omplemenary o he Chinese speializaion paern and orhogonal o he Indian one. Figures 5, 6, 7 and 8 show he orrelaion beween Andean ounries, Cenral Ameria, Mexio, and he Souhern Cone, on he one hand, and China and India on he oher. Again, Mexio is no inluded in he Cenral Amerian aggregae, beause is RCAs and he orrelaion of is RCA is very differen from he ones observed for he res Cenral Amerian and LAC. In he ase of he Andean ounries (Figure 5) he paern is very similar o he one observed for LAC as a whole. In he ase of Cenral Ameria (Figure 6) he orrelaion wih China shows a similar rend, bu he orrelaion wih India remains relaively sable and posiive, suggesing ha India had a speializaion paern ha evolved in he same direion as Cenral Ameria s rade paern. For Mexio, he paern is quie differen. The rend is upwards in boh ases sine he early 1990s, suggesing ha Chinese and Indian rade is beoming more similar o Mexio s, bu we anno oversae he fa ha his finding does no neessarily mean ha Chinese and Indian expors are huring Mexian expor prospes, sine boh ould be rising ogeher perhaps as a resul of he operaion of global produion neworks. In he ase of he Souhern Cone, boh orrelaions show a delining rend, whih suggess ha he Souhern Cone is moving away from he speializaion paern of China and India. Noe, however, ha in he ase of India, he orrelaion is sill posiive and relaively high by he end of he period (around 0.2). Thus, wih he exepion of Mexio, China and India seem o be speializing in a differen direion when ompared o LAC, and heir rade paerns are beoming more omplemenary suggesing ha he growh of China and India may be good news for LAC wih he exepion of Mexio Why is he orrelaion of LAC s RCAs wih China and India s RCA falling? The fa ha he speializaion paern of LAC is moving away from he speializaion paern of China and India may be due o differen fores. On he one hand, LAC may be 9

11 speializing in a few produs (onenraing is expors) where China s presene is no very srong or delining. On he oher, LAC may be diversifying is expor bundle ino new seors. In order o disenangle wha s driving he evoluion of he orrelaion of RCAs, Figures 4 o 8 also show he evoluion of an expor onenraion Herfindhal index (higher values indiae a more onenraed expor bundle), where he righ hand side verial axis provides he sale, and he green line he evoluion of he index. The evidene suggess ha LAC as a whole has been moving owards higher onenraion of is expor bundle hroughou he period. This rend is mainly driven by a srong move owards onenraion of expors by he group of Andean ounries sine he mid 1990s and some mild rend owards onenraion in he Souhern Cone. In onras, Cenral Ameria has shown some srong diversifiaion of is expor bundle, and Mexio some mild diversifiaion. During he same period China moved owards a more onenraed expor bundle, in pariular sine he mid 1990s, whereas India has shown some diversifiaion. Overall his suggess ha he explanaion behind he falling orrelaion beween LAC and China is ha LAC and China are moving owards more speializaion bu in a differen se of produs. In he ase of India, he rend would also be explained by he diversifiaion of India s expor bundle. Some key quesions remain unanswered. For example, in whih seors is LAC speializing and wha are he seors in whih China and India are speializing? In order o answer hese quesions we look a he evoluion of RCAs beween 1990 and 2004 a he 3 digi level of he ISIC for LAC, he Andean group, Cenral Ameria, Mexio, he Souhern Cone, China and India. In order o fous on he observed negaive rend in he orrelaion beween RCAs of LAC and hose of China and India, we idenify 5 poenial ases whih an explain his rend: 1. China/India RCA inreases, and LAC (or any of he LAC sub-groups) RCA remains onsan (China/India is speializing in a produ where LAC omparaive advanage remains sable). LAC is beoming relaively less ompeiive (or a larger imporer of his produ). 10

12 2. LAC RCA inreases, and China/India RCA remains onsan (LAC is speializing in a produ where China/India s omparaive advanage remains sable). LAC is beoming relaively more ompeiive. 3. China/India RCA inreases, and LAC RCA delines (China/India is speializing in a produ where LAC is wihdrawing). China/India is a hrea for LAC. 4. LAC RCA inreases, and China/India RCA delines (LAC is speializing in a produ where China/India is wihdrawing as an exporer). LAC is aking he opporuniy offered by China/India s wihdrawal as an exporer. 5. China/India RCA delines, and LAC RCA remains onsan (China/India is wihdrawing as an exporer from a produ where LAC s omparaive advanage remains sable). China/India s wihdrawal as an exporer does no seem o have an effe on LAC. 6. LAC RCA delines, and China/India RCA remains onsan (LAC is wihdrawing as an exporer from a produ where China/India s omparaive advanage remains sable). LAC s wihdrawal as an exporer does no seem o be aused by China/India. Explanaions 2 and 4 are probably good news for LAC, whereas explanaions 1 and 3 are probably bad news. Explanaions 5 and 6 are neural. We hen aegorize eah indusry aording o one of hese six aegories ha ould poenially explain he fa ha LAC and China and India s speializaion paerns are moving apar and beoming more omplemenary. Indusries where he observed RCA rends anno help explain he negaive rend in RCAs orrelaions are aegorized wih a value of 0. These are indusries ha ould eiher explain a posiive or a zero rend in he orrelaion beween LAC RCAs and he RCAs of China and India. Table 2 provides he full aegorizaion by indusry for LAC as a whole and for eah of he sub-groups. The evidene suggess ha he rend is mainly explained by he evoluion of RCAs in agriulure, fishing, foresry and mining. Bu he explanaion has more o do wih a delining omparaive advanage of China, due o is growing demand in hese 11

13 seors, wihou observing an inrease in he omparaive advanage of LAC. I seems ha in hese seors LAC is parly missing he opporuniy offered by he growh in demand in China and India. Generally, he negaive rend anno be explained by he evoluion of RCAs in he manufauring seor (mosly 0 ). There are some exepions: food manufauring and beverages, where LAC inreased is RCA whereas China s and India s RCA delined (parly due o heir growing demand); professional and sienifi equipmen, where China s RCA is delining; obao, exiles and iron and seel where China s RCA is rising and LAC s is delining; prining, paper, poery, non-meali mineral produs and fabriaed meal produs where eiher China or India experiened rising RCAs, bu wihou any disernable rend in he RCA of LAC in hese seors. There is lile heerogeneiy aross LAC sub-groups wih he exepion of Mexio where 0 are more ommon. The evoluion of Mexian, Chinese, and Indian RCAs by indusry anno explain a negaive rend in he orrelaion of RCAs beween Mexio and China and India. This is no surprising given ha Mexio was he only sub-group for whih we observed a posiive rend in he orrelaion of RCA wih China and India s RCAs. Thus, he good news is ha LAC s speializaion paern seems o be moving away from ha in China and India, leading o more omplemenary rade speializaion paerns. However, his is mainly due o he evoluion of heir relaive RCA in agriulure, fisheries, foresry and mining, bu no in manufauring (wih a few exepions), where RCAs of China and India are moving in he same direion as he RCAs of LAC ounries (wih a few exepions suh as food manufauring, beverages, and professional and sienifi equipmen). The bad news is ha LAC eonomies do no seem o be aking advanage of he growing appeie of China for raw maerials, as hey mainain he same level of RCA as China s omparaive advanage delines. 9 Thus, as world markes for hese produs inrease, LAC is no able o grab a bigger share of his growing pie. 9 Noe ha beause he size of hese markes is growing, his is no all bad news for LAC. 12

14 3. Where are China and India pushing LAC s speializaion paern? The objeive of his seion is hree fold. Firs, we explore wih a more sysemai approah and onrolling for ounry*year effes wheher China and India RCAs are moving ogeher or in opposie direions. Seond, we explore he role played by heir bilaeral rade on heir overall speializaion paern. Finally, we idenify he broad aegories of produs defined aording o heir faor use ha are experiening sronger ompeiion or demand from China and India. Are unskilled inensive seors being hur by he enhaned speializaion of China and India in hese produs? Are naural resoure seors experiening sronger demand from China and India han oher seors? Wha abou skilled-inensive seors and seors inensive in sienifi knowledge? The answers o hese quesions would help arge he orre faor of produions when designing poliies o help maximize he opporuniies offered by he growh of China and India, while minimizing he adjusmen oss. 3.2 Empirial Mehodology The empirial mehodology is sraighforward. We explain he RCAs of LAC eonomies wih he RCAs of China and India, as well as he bilaeral expors of eah LAC eonomy wih China and India, onrolling for ounry*year fixed effes. Tha is, our empirial model an be wrien as: RCA LAC, s, = β + β 0 + α XN 4 LAC, + α RCA 1 INDIA LAC, s, CHINA, s, + ε LAC, s, + α RCA 2 INDIA, s, + α XN 3 CHINA LAC, s, (5) where RCA, LAC, s is he RCA of ounry (belonging o our hireen LAC ounries) in seor s, a ime, XN are ne bilaeral expors of eah LAC eonomy o eiher China or India depending on he variable, and ε is an error erm where we allow for LAC, s, lusering of he error erm wihin eah indusry every year. We esimaed hese models for he pooled sample of hireen LAC ounries, bu also for he 4 sub-groups (Andean, Cenral Ameria, Souhern Cone, and Mexio). 13

15 A posiive oeffiien on he RCA of China or India would indiae ha LAC s speializaion paern is similar o he one observed in China and India, whereas a negaive oeffiien would indiae ha he speializaion paern of LAC is omplemenary o he speializaion paern of China and India. A posiive oeffiien on he bilaeral ne expor variable would indiae ha expors o China or India are onenraed in seors where LAC s omparaive advanage lies, and ha a leas hrough his dire hannel he growh of China and India is shaping he speializaion of LAC eonomies. In order o see he faors of produion ha may be hur or helped by inreased ompeiion from China and India, we also esimaed equaion (5) wih dummies for unskilled labor inensive seors, skilled labor inensive seors, naural resoure inensive seors, and sienifi knowledge inensive seors, as well as heir ineraion wih he RCAs of China and India. The inerpreaion of he oeffiiens on he ineraion variables is similar o he one disussed above. One hallenge for his approah is ha researhers need o define he seoral faor inensiies. There are broad lassifiaions for some of hese faors, bu hey are no speifi o LAC. A he level of aggregaion a whih we work, seors ha are unskilled inensive in OECD ounries may be skilled inensive in LAC (e.g., exiles) as disussed by Feensra and Hanson (2003). Moreover, o our knowledge here is no worldwide lassifiaion of faor-inensiy a he 3 digi level of he ISIC (our level of disaggregaion). Consequenly here is no oher soluion bu o esimae he faor inensiies of eah seor. To do his, we follow Kohli (1991), Harrigan (1997), and Redding (2002) revenue funion approah and esimae he sign and saisial signifiane of he Rybynski elasiiies for eah of he four faors of produion in eah of he ISIC 3 digi indusries for he pooled sample of hireen LAC ounries. The full mehodology used o esimae he faor inensiies and he resuls are disussed in he appendix. 3.3 Resuls 14

16 The ondiional orrelaion of LAC and eah of he four sub-regions RCAs wih he RCAs of he wo Asian eonomies esimaed following equaion (3) are shown in Table 3. Looking a he firs olumn of Table 3, LAC s RCA is negaively orrelaed wih China s RCA, alhough he oeffiien is no saisially signifian. This suggess ha China s RCA is a wors no orrelaed wih LAC s RCA, and herefore one shouldn expe muh ompeiion from China in world markes, bu raher a mild reinforemen of LAC s RCA on produs due o larger Chinese demand. On he oher hand, he orrelaion wih India s RCA is muh larger, posiive and saisially signifian, suggesing ha ompeiion from India in world markes may be more of a hallenge for LAC s eonomies. As India s size in world markes is only abou a fifh of China s size, his may be no be a problem oday, bu raher in he fuure if India oninues o grow a is urren raes, whereby in less han weny years India would have he same size as China oday. Bilaeral ne expors are insignifian in all regressions, suggesing ha LAC s expors o China and India are neiher posiive nor negaively orrelaed wih LAC s omparaive advanage one we onrol for China and India s RCAs (and ounry*year dummies). The fa ha we are onrolling for China s RCA and ha he Chinese RCA ends o be negaively orrelaed wih Lain Ameria s RCAs suggess ha he insignifian oeffiien on bilaeral ne expors of LAC o China may be due o he fa ha hese variable is orrelaed wih China s RCA, i.e., he bilaeral ne expors of LAC o China are orrelaed wih he overall Chinese RCA. I also refles he relaively small bilaeral rade beween LAC and he wo Asian eonomies. There are some ineresing differenes aross regions. China s RCA has a posiive and signifian oeffiien for he sub-sample of Andean ounries. The oeffiien is also posiive for Mexio, bu saisially insignifian. These posiive oeffiiens sugges ha for hese ounries he Chinese ompeiion in hird markes may be sronger han for LAC as a whole. In he ase of Cenral Ameria and he Souhern Cone he oeffiien on he Chinese RCA is negaive and signifian, suggesing ha he Chinese speializaion paern may be more omplemenary o he speializaion paern in hese wo regions. In 15

17 onras, he speializaion paerns of hese wo regions seem o be more similar o he speializaion paern of India wih large posiive and signifian oeffiiens on India s RCA. The oeffiien on he Indian RCA is also posiive for Mexio, bu he oeffiien is saisially insignifian. In he ase of he Andean ounries, he oeffiien on he Indian RCA is small, bu negaive and signifian suggesing ha he rade speializaion paern of he Andean ounries ends o be a bi more omplemenary wih he speializaion paern of India han he one of LAC or Mexio. Table 4 provides he resuls for he speifiaions where we inrodue in equaion (5) he ineraion erms beween China and India RCA and he four faor inensiy dummies (unskilled labor, skilled labor, naural resoures and sienifi knowledge), as well as he four faor inensiy dummies separaely. The oeffiiens on he dummies sugges ha on average hroughou he period LAC s omparaive advanage lies in seors ha are inensive in naural resoures, sienifi knowledge and unskilled labor, in dereasing order. Is omparaive disadvanage lies in skilled-labor inensive seors. There is some heerogeneiy aross regions, however. Andean ounries faor inensiies refle he one observed for LAC as a whole. Cenral Ameria has is omparaive advanage in seors inensive in naural resoures and unskilled labor (alhough he laer is no saisially signifian), and hey have a omparaive disadvanage in seors inensive in sienifi knowledge and skilled labor. Mexio s omparaive advanage seems o lie in unskilled- and skilled-labor inensive seors, and naural resoures, whereas is omparaive disadvanage lies in seors inensive in sienifi knowledge (alhough none of he oeffiiens are saisially signifian). The Souhern Cone omparaive advanage lies in naural resoure inensive seors, skilled labor and sienifi knowledge, whereas is omparaive disadvanage lies in unskilled inensive seors. Whih faors will be more affeed by he growh of China and India in world markes? The oeffiiens on he ineraion erms repored in Table 4 allow us o answer his quesion. In he ase of he pooled sample for he hireen LAC ounries all he ineraion erms are saisially signifian wih he exepion of he ineraion of China s RCA wih he dummies for seor inensiy in unskilled labor and sienifi 16

18 knowledge. Seors inensive in unskilled labor would suffer from he rising ompeiion from India in world markes, bu hese seors show some small (and saisially insignifian) omplemenariy wih he speializaion paern of China. Seors ha are inensive in he use of skilled workers would be suffering from he rising ompeiion from India and China in world markes, alhough i is worh noing ha hese indusries may be unskilled-labor inensive in China and India (as seors faor inensiies were esimaed in a sample ha only inluded LAC ounries). Seors inensive in sienifi knowledge show srong omplemenariy wih India s speializaion paern. Finally, naural resoure inensive seors show some srong omplemenariies wih he speializaion paern of boh China and India. There are some mild differenes aross LAC regions. For example, in Cenral Ameria and he Souhern Cone omplemenariies seem o be onenraed in naural-resoure inensive seors, and no sienifi knowledge inensive seors. On he oher hand in Mexio omplemenariies are onenraed in sienifi-knowledge inensive seors. There are no large qualiaive differenes beween LAC s resuls and he resuls for eah of he sub-regions. However, overall hese resuls sugges ha he growh of China and India is more likely o affe negaively seors ha are relaively labor inensive (skilled and unskilled), whereas naural-resoure and sienifi-knowledge inensive seors in LAC are likely o benefi from China and India s growh in world markes. 4. Conluding Remarks Is LAC ompeing in he same produs wih China and India s exporers, or is he growing demand of China and India in world markes helping Lain Amerian exporers? We answer hese quesions by exploring he orrelaion beween he rade speializaion paerns of Lain Amerian eonomies on he one hand, and he speializaion paerns of China and India on he oher. Resuls sugges ha overall he speializaion paern of LAC eonomies wih he exepion of Mexio has been moving in opposie direion o he rade speializaion paern of China and India. This indiaes ha LAC s rade speializaion paern is 17

19 beoming more omplemenary o he speializaion paern of China and India. China s speializaion paern a he end of he period sudied (early 2000s) was negaively orrelaed wih he speializaion paern of mos LAC eonomies (again Mexio being an exepion). India s speializaion paern, however, seems o be posiively orrelaed even a he end of he period wih hose of mos LAC eonomies (wih he exepion of Andean ounries). However, given he smaller size of India (i represens a fifh of China in world markes), his may no be bad news so far. Regarding he poenial effes ha China and India as expor markes for LAC may have on he speializaion paerns of LAC eonomies, we found no evidene of signifian effes. The bilaeral ne expors of LAC o he wo Asian eonomies do no seem o have a disernable impa on LAC s speializaion paern in spie of heir rapid growh, probably due o he sill relaively small size of he bilaeral rade of LAC eonomies wih China and India. Joinly hey represen 5 peren of LAC s expors and 7 peren of LAC impors. Finally, regarding he faors of produion ha are likely o be affeed by he growing presene of China and India in world markes, we found ha labor (boh unskilled and skilled) was he faor ha was more likely o be negaively affeed, whereas naural resoures and sienifi knowledge were likely o benefi from he growing presene of China and India in world markes. LAC governmens may need o envisage poliies aiming a miigaing he poenial adverse effe of he growh of China and India on unskilled and skilled labor by argeing workers in he affeed indusries, raher han by raising proeionis barriers o rade sine he laer will also hur domesi onsumers, and users of impored inermediae goods, whih will redue he poenial expor gains by seors ha are benefiing from he emergene of China and India. In his regard, his paper onribues o his poliy disussion by idenifying he speifi seors ha may be shedding skilled and unskilled workers as a onsequene of he growh of he wo Asian eonomies. 18

20 Referenes Barro, Rober J. and Jong-Wha Lee (2000), Inernaional Daa on Eduaional Aainmen: Updaes and Impliaions, CID Working Paper No. 42. Blazquez-Lidoy, J., J. Rodriguez and J. Saniso (2006), Angel or Devil? Chinese rade impa on Lain Amerian Emerging Markes, Working Paper #252, OECD. Calderon, C. (2006), Trade, Speializaion and Cyle Synhronizaion: Explaining Oupu Co-movemen beween Lain Ameria, China and India, mimeo, The World Bank. Casro, L., Casro, J. Tramuola and L. Mona (2005) China: omo puede la Argenina aprovehar la gran oporunidad, Edhasa Ediorial, Buenos Aires. Feensra, R. and G. Hanson (2003), Global Produion Sharing and Rising Inequaliy: A Survey of Trade and Wages, in Choi E.K. and J. Harrigan (eds.) Handbook of Inernaional Trade. Blakwell: Malden, MA, Harrigan, J., (1997), Tehnology, Faor Supplies, and Inernaional Speialisaion: Esimaing he Neolassial Model, Amerian Eonomi Review 87, Kohli, U., (1991), Tehnology, Dualiy and Foreign Trade Ann Arbor, Universiy of Mihigan Press. Lederman, D., M. Olarreaga and I. Soloaga (2006), The growh of China and India in world markes: opporuniy or hrea for Lain Amerian exporers?, mimeo, The World Bank. Niia, A. and M. Olarreaga (2006), Trade, Produion and Proeion, , available a 19

21 Redding, S. (2002), Speializaion Dynamis Journal of Inernaional Eonomis 58, Vollrah, T. L (1991), A heoreial evaluaion of alernaive rade inensiy measures of revealed omparaive advanage, Welwirshaflihes Arhiv 130,

22 Appendix: Deermining faor inensiies by indusry in Lain Ameria In order o deermine indusries faor inensiy, we will esimae Rybynski elasiiies for four faor of produion (unskilled labor, skilled labor, sienifi knowledge, and naural resoures). Following Kohli (1991), Harrigan (1997) and Redding (2002) we assume a ranslog GDP funion whih depends on goods pries, faor endowmens and oal faor produiviy. The ranslog funion is suffiien flexible o loally approximae he rue underlying revenue funion quie losely. We assume he GDP funion is ommon aross all ounries (he hireen Lain Amerian eonomies in our sample) and ime (exep for ommon Hiks neural ehnology differenes aross ime). Imposing he symmery of ross effes, assuming linear homogeneiy of degree 1 in faor endowmens and pries, aking he firs order ondiion and rearranging we obain a share equaion where he onribuion of eah indusry in GDP is explained by he log of all goods pries, he log of all faors of produion and he log of he ommon TFP. We furher assume ha goods are homogenous and raded freely so ha all goods pries are equal. This allows us o replae all pries by year dummies, whih will also apure he ommon TFP. Thus, he final share equaion is explained by he log of faor endowmens and year dummies, allowing us o fous on he impa ha an inrease in eah faor endowmen would have on he level of produion in eah indusry in Lain Ameria. The equaion o esimae for eah indusry is: s = α ln( unskilled ) + β ln( skilled ) + γ ln( sienifi_ knowledge ) + λ ln( naural_ resoures ) + δ + ε (A.1) where s = value _ added GDP is he share of eah indusry s in he ounry s GDP. The oher variables are he per apia faor endowmens by ounry and year. Skilled and unskilled endowmens ome from from he Barro and Lee (2000) daase on eduaion. Unskilled individuals are defined as hose wih inomplee seondary eduaion. Skilled individuals are hose wih omplee seondary eduaion or higher. Sienifi knowledge is proxied by he number of sienifi journals per apia, and naural resoures by arable 21

23 land. Daa for hese wo oher endowmens and GDP omes from he World Developmen Indiaors. Daa on value added by indusry omes from UNIDO. Regressions are run for he hireen Lain Amerian eonomies in our sample. The ime overage is poenially from 1988 o 2004, bu his depends on daa availabiliy (in pariular value-added). For a desripion of daa availabiliy for rade and value-added see Niia and Olarreaga (2006). We will deermine ha a good is inensive in one of he four faors above if he oeffiien in fron of ha faor is posiive and signifian a leas a he 10 peren level. Appendix Table 1 shows when his ours in eah 3-digi ISIC indusry, by providing he esimaed oeffiien for hose variables ha are posiive and signifian in eah regression. Noe ha some indusries may be inensive in several faors of produion. For example, agriulure is inensive in naural resoures and unskilled labor, whereas mining is inensive in unskilled labor and sienifi knowledge. On he oher hand, for foowear and oher manufauring indusries ould no esimae heir faor-inensiies. We also alulaed RCAs by faor inensiy for he years 1990 and 2004 for LAC as a whole. Resuls are repored in Appendix Table In 1990 LAC had a srong omparaive advanage in naural resoure inensive seors, and a srong omparaive disadvanage in unskilled labor. There has been lile movemen by 2004, wih a sligh improvemen in he omparaive advanage of seors inensive in unskilled labor and sienifi knowledge, and a deline in he omparaive advanage in naural resoures and skilled labor. The are small differenes wihin he region, wih he exepion of Mexio ha has a omparaive advanage in sienifi-knowledge inensive seors and a omparaive disadvanage in skilled-inensive seors. 10 One should noe ha here is some double-ouning involved as, for example, some seors are inensive in boh unskilled labor and naural resoures. 22

24 Appendix Table 1: Esimaing faor inensiies in LAC indusries ISIC CODE INDUSTRY NAME 311 Food produion 313 Beverage indusries 314 Tobao manufaures 321 Manufaure of Texiles UNSKILLED LABOR (0.349) 3.59 (0.555) SKILLED LABOR FACTOR INTENSIVE NATURAL RESOURCES SCIENTIFIC KNWOLEDGE 2 R (0.144) (0.191) (0.116) (0.177) Wearing apparel Manufaure of Leaher (0.17) (0.052) Manufaure of Foowear Manufaure of Wood (0.033) * (0.028) Manufaure of Furniure (0.113) (0.010) Manufaure of Paper (0.044) Prining, publishing (0.200) (0.0642) (0.0186) Indusrial Chemials (0.968) 0.532* (0.289) (0.093) Oher hemial produs (0.707) (0.066) Peroleum refineries 1.546* (0.830) (0.266) Misellaneous produs Rubber produs (0.151) (0.013) Plasi produs (0.021) Manufaure of Poery (0.087) (0.022) Manufaure of Glass (0.009) Oher non mealli produs (0.03) Iron and seel basi indusries (0.044) Non-ferrous meals (0.166) Fabriaed meal prod (0.036) Mahinery exep elerial (0.320) (0.029) Elerial mahinery (0.371) (0.036) Transpor equipmen (0.631) (0.0579) Professional and sienifi (0.054) (0.018) Oher manufauring indusries Agriulure (33.508) (9.247) Mining 1.277* (0.727) (0.144) Noe: We only repor oeffiiens ha were posiive and saisially signifian a he 10 peren level. Sandard errors are provided in parenheses. * Denoes signifiane a he 10%; for all oher repored oeffiiens heir saisial signifiane is higher. 23

25 Appendix Table 2: RCAs by faor, LAC Andean Cenral Ameria Souhern Cone Mexio Unskilled Skilled High Teh Naural Resoures Noe: Auhors alulaions as desribed in he ex of he appendix. 24

26 Figure 1: Share of China and India in Lain Amerian expors, 1990 versus 2004 Peru Chile Argenina Brazil Uruguay LAC Niaragua Cosa Ria Bolivia Colombia Euador Mexio Guaemala Venezuela Soure: Unied Naions' Comrade Figure 2: Share of China and India in Lain Amerian impors, 1990 versus 2004 Chile Euador Peru Mexio Colombia Brazil LAC Bolivia Uruguay Niaragua Argenina Cosa Ria Venezuela Guaemala Soure: Unied Naions' Comrade

27 Figure 3: The growing Chinese appeie for Commodiies Share in World Consumpion Tin Zin Soy bean Aluminium Copper Nikel Sugar Peroleum Figure 4: Is LAC ompeing in he same produs as China and India? Correlaion Index Year Corr(LAC, China) Herfindhal LAC Corr(LAC, India) Soure: Auhors' alulaions 26

28 Figure 5: Are Andean ounries ompeing in he same produs as China and India? Correlaion Index Year Corr(Andean, China) Herfindhal Andean Corr(Andean, India) Soure: Auhors' alulaions Figure 6: Is Cenral Ameria ompeing in he same produs as China and India? Correlaion Index Year Corr(C.Ameria, China) Herfindhal Cenral Ameria Corr(C.Ameria, India) Soure: Auhors' alulaions 27

29 Figure 7: Is Mexio ompeing in he same produs as China and India? Correlaion Index Year Corr(Mexio, China) Herfindhal Mexio Corr(Mexio, India) Soure: Auhors' alulaions Figure 8: Is The Souhern Cone ompeing in he same produs as China and India? Correlaion Index Year Corr(S.Cone, China) Herfindhal Souhern Cone Corr(S.Cone, India) Soure: Auhors' alulaions 28

30 Table 1. RCA indies by seor and ounry/region in 1990 and 2004 LAC Andean Cenral Ameria Souhern Cone Mexio China India ISIC Name Agriulure Logging Fishing Coal Mining Crude per Ore mining Oher mining Food manuf Texile&App Wood prod Paper prod Chemials Non-meal prod Basi meal ind Fabriaed meal Oher manuf Eleriiy&gas NA NA NA Noe: When he RCA akes he value of -, his indiaes ha he ounry did no expor ha produ in ha year, bu ha i impored some, when i akes he value of +, i indiaes ha he ounry did no impor ha produ in ha year, bu ha i expored some; NA indiaes ha he ounry neiher impor, nor expor ha produ in ha year. Numbers in bold indiae ha here was an inrease of he RCA of ha indusry in ha ounry/region of more han 0.5 poins beween 1990 and 2004, and numbers ha are undersore indiae ha here was a deline of he RCA of ha indusry in ha ounry/region of more han 0.5 poin beween 1990 and All oher indusry-ounry ombinaions are wihin he -0.5 and +0.5 poin hange in he RCA. 29