Trde Liberlizion nd Expor Vriey: A Comprison of Mexio nd Chin Rober C. Feensr Hiu Looi Kee Running Tile: Trde Liberlizion nd Expor Vriey ROBERT C. FEENSTRA, Deprmen of Eonomis, Universiy of Cliforni, Dvis, nd NBER; HIAU LOOI KEE, The World Bnk. Reserh funding from he World Bnk is grefully knowledged. The findings, inerpreions, nd onlusions expressed in his pper re enirely hose of he uhors, nd do no neessrily represen he view of he World Bnk, is Exeuive Direors, or he ounries hey represen. The uhors hnks he nonymous referees for helpful ommens.
2 1. INTRODUCTION The hllmrk of he endogenous growh models (Romer 1990, Grossmn nd Helpmn, 1991) is heir fous on he reion of new or higher quliy produs, nd he effes of suh innovions on produiviy nd eonomi growh. Opening ounry o rde opporuniies due o riff reduions will ypilly inrese he produ vriey of impors vilble, nd my lso inrese he vriey of expors, boh of whih onribue o growh. Despie he miroeonomi fous of hese models, he link beween rde nd growh is usully ssessed more ggrege level, in whih se he usliy beween he wo is unler (Frnkel nd Romer, 1999, Dollr nd Kry, 2001, Rodriguez nd Rodrik, 2000). To move beyond hese ggrege sisis, we need more deiled informion on he produ vriey of rded goods, nd on he link beween riff reduions nd produ vriey. The issue of mesuring produ vriey hs reeived relively lile enion due o is inheren diffiuly. In he lnguge of index numbers, n expnsion in he rnge of inpus or oupus is new goods problem: good h is newly vilble will hve n observed prie nd quniy, bu no orresponding prie or quniy he yer before. The vilbiliy of his new good will yield welfre gin o onsumers, s well s produiviy gin o firms buying he new inpu. In his pper we show how produ vriey n be mesured in he se of CES ggregor funion. We pply hese resuls o he mesuremen of expor vriey from Chin nd Mexio o he US. The ppliion o Chin nd Mexio is moived by he hnges in rde poliy fing hose ounries in reen yers. Mexio joined he Norh Ameri Free Trde Are (NAFTA) in 1994, whih subsnilly lowered i riffs o he US nd Cndin mrkes. Wh hs hppened o Mexio s expor vriey sine h ime? Kehoe nd Ruhl (2002) rgue h goods from Mexio
3 h were he les rded before riff liberlizion oun for disproporione growh in rde following he reduion of rde brriers. We will lso doumen he expnsion in expor vrieies from Mexio due o NAFTA. Similrly, while Chin ws eped ino he WTO in 2000, i ws implemening unilerl riffs reduions of is own before h ime nd benefiing from low riffs brod. We will invesige he growh in expor vriey from Chin over 1990-2001, nd ompre hose findings o Mexio. In ddiion, we will rgue h he expnsion of Chin s expor vriey due o he fll of US riffs hs used n dverse mrke ompeiion effe on he expor vriey from Mexio. This pper is orgnized s follows. Afer reviewing he lierure on he new goods problem in seion 2, we disuss how o mesure expor vriey in he CES se in seion 3. In seions 4 nd 5 we disuss he empiril ppliions o expor vriey growh in Mexio nd Chin. Regression resuls reling rde liberlizion o indusry expor vriey re presened in seion 6, nd onlusions re given in seion 7. 2. LITERATURE REVIEW. Theoreil Resuls The problem of ompuing welfre gins for onsumer due o new goods is very similr o evluing he produiviy gins for firm wih new inpus. Hik s (1940) reommended one soluion o his problem: newly vilble good should be evlued is reservion prie when i is no vilble, where demnd is zero. When he new good beomes vilble hen demnd is posiive nd so is prie is lower hn he reservion prie. Thus, he fll in pries n be ompued s he differene beween he reservion nd observed pries, nd inegring he demnd urve beween hese pries is mesure of he onsumer welfre gin, or firm produiviy gin, due o he new good. Exmples of his pproh pplied empirilly inlude
4 Grilihes nd Cokburn (1994) nd Bernd e l. (2002) for generi drugs, nd Husmn (1997, 1999) for brekfs erels nd ellulr elephones. The produiviy gin from new inpu for firm is illusred in Figure 1. Given oupu y = y, he inpus would lie long n isoqun ACD like h illusred. If only inpu 1 is vilble, hen he oss of produing y would be minimized poin A, wih he budge line AB. Bu if inpu 2 is lso vilble, hen he oss re insed minimized poin C, wih fll in oss. This illusres he benefis of inpu vriey. The diffiuly wih pplying his soluion in prie is h reservion pries re no h esy o esime, espeilly when here re mny new goods ppering. A simpler soluion is proposed by Feensr (1994) for h se where new goods pper wihin CES ggregor funion (i.e. produion funion for firms, expendiure funion for onsumers, or produion possibiliy fronier for n eonomy). Suppose h he elsiiy of subsiuion beween he goods is σ > 1. In h se, he reservion prie for ny good is infinie: he isoquns of he firm hi he xis in Figure 1 wih slope zero ( poin A) nd infiniy ( poin D). Bu he ol reduion in oss due o he new inpu n sill be ompued. Suppose h he se of inpus vilble o he firm eh period is I, = 0,1, wih he ommon se I ( I 0 I ) 1 oss fll due o he pperne of new inpus by he moun:. Feensr (1994) shows h λ1 λ0 ( I ) ( I ) 1/ ( σ 1), (1) where he vlues λ (I) re onsrued s: p i I i xi p i I i I i x, i λ ( I ) = 1, = 0, 1. (2) p i I i xi p i I i xi
5 In hese expressions, I denoes he se of inpus vilble in periods = 0, 1, he pries p i nd wih os-minimizing quniies x i. New goods will be in he se I 1 bu no I, wheres disppering goods re in he se I 0 bu no I. From (2), eh of he erms λ (I) 1 n be inerpreed s he period expendiure on he goods in he se I relive o ol expendiure in h period. Alernively, his n be inerpreed s one minus he shre of period expendiure on new goods (no in he se I). When here is greer number of new goods in period, his will lower he vlue of λ (I). Noie h he rio [λ 1 (I)/ λ 0 (I)] in (1) is rised o he power 1/( σ 1), so wih σ > 1 lower vlue of λ 1 (I) due o new inpus will redue he rio in (1), by n moun depending on he elsiiy of subsiuion. While (1) mesures he reduion in oss due o he pperne of new inpus, he erm [λ 1 (I)/ λ 0 (I)] iself is n inverse mesure of produ vriey: when [λ 1 (I)/ λ 0 (I)] < 1 hen here re more new hn disppering goods, nd produ vriey is expnding. By simply invering his erm, we obin dire mesure of produ vriey h n be implemened using d on observed expendiures on goods. There re severl ppliions of his mehod o mesuring he vriey of rded goods, s disussed in he nex seion. Before urning o hese ppliions, noe h he formuls in (1) nd (2) nno be pplied when 0 σ 1, beuse in h se inpus re essenil o he produion proess, nd hving zero of ny inpu resuls in zero oupu. So we nno even onsider new goods in h se. However, hese formuls re sill relevn when σ < 0. Th se pplies o mesuring he benefis from oupu vriey for n eonomy. This is illusred in Figure 2, where we show he produion possibiliy fronier (PPF) beween wo oupu vrieies x 1 nd x 2. For given PPF, nd given pries, n inrese in he number of oupu vrieies will rise revenue. For exmple, if only
6 oupu vriey 1 is vilble, hen he eonomy would be produing he orner A, wih revenue shown by he line AB. Then if vriey 2 beomes vilble, he new equilibrium will be poin C, wih n inrese in revenue. When σ < 0, hen (1) mesures he inrese in revenue due o he pperne of new oupus, nd he rio [λ 1 (I)/ λ 0 (I)] is n inverse mesure of oupu vriey. b. Empiril Appliions There re severl sudies h mesure he benefis of inpu or oupu vriey in rising produiviy. Feensr e l. (1999) provide n ppliion of his mehod o indusry produiviy growh in Souh Kore nd Tiwn. The d used o mesure vriey re he disggrege expors from hese ounries o he Unied Ses. Those uhors nlyze he relionship beween hnges in expor vriey nd he growh in TFP ross Souh Kore nd Tiwn, for sixeen seors over 1975-1991. They find h hnges in relive expor vriey (enered s eiher lg or led) hve posiive nd signifin effe on TFP in nine of he sixeen seors. Seven of hese seors re lssified s seondry indusries, in h hey rely on s well s produe differenied mnufures, nd herefore seem o fi he ide of endogenous growh. Among he primry indusries, whih rely more hevily on nurl resoures, he uhors find mixed evidene: he orrelion beween expor vriey nd produiviy n be posiive, negive, or insignifin. In ddiion, he uhors lso find evidene of posiive nd signifin orrelion beween upsrem expor vriey nd produiviy in six downsrem seors, five of whih re seondry indusries. Funke nd Ruhwedel (2001) hve pplied he sme mesure of produ vriey o nlyze eonomi growh ross he OECD ounries. Using pnel dse of 19 ounries over 1989-1996, hey find h ounry s expor vriey relive o he US is signifin deerminn of is GDP per-pi. Noie h hese mesures of produ vriey, whih re onsrued from highly disggrege rde d, re unlikely o suffer from he endogeneiy problem h plgues
7 ggrege rde flows (s disussed by Frnkel nd Romer, 1999). Therefore, he onsruion of produ vriey indexes nd heir orrelion wih TFP offers n lernive wy o ssess he imporne of rde in eonomi growh. More reenly, Hummels nd Klenow (2005) deompose he growh of world rde ino h pr due o ounries exporing new produs wh hey ll he exensive mrgin nd h pr due o ounries exporing more of he sme produs he inensive mrgin. They find h exensive mrgin ouns for wo-hirds of he greer expors of lrger eonomies, nd onehird of heir impors. In noher ppliion, Brod nd Weinsein (2005) mesure he imp on welfre for he imporer. For he Unied Ses, hey find h he upwrd bis in he onvenionl impor prie index (due o ignoring produ vriey) is pproximely 1.2 per en per yer, implying h he welfre gins from umulive vriey growh in impors re 2.8 per en of GDP in 2001. Finlly, Feensr nd Kee (2006) esime he imp of expor vriey on produiviy growh for group of ounries, nd Brod, Greenfield nd Weinsein (2006) show how impor vriey is reled o ounry produiviy. 3. MEASURING EXPORT VARIETY Our resuls bove hve been sed in erms of hnges in produ vriey over ime. Bu he sme resuls pply o omprison of wo ounries poin in ime. Suppose h he se of expors from ounries nd differ, bu hve some produ vrieies in ommon. Denoe his ommon se by I ( I I ) from ounry relive o ounry is: λ λ. Rewriing (2), n inverse mesure of expor vriey ( I ) ( I ) where, ( I ) piqi λ i I p q. (3) i I i i i
8 Noie h λ ( I ) 1 in (3) due o he differing summions in he numeror nd denominor. This erm will be srily less hn one if here re goods in he se I h re no found in he ommon se I. In oher words, if ounry is selling some goods in period h re no sold by ounry, his will mke λ ( I ) < 1, so i is n inverse mesure of ounry expor vriey. The rio [ λ ( I ) / λ ( I )] is n inverse mesure of expor vriey from ounry relive o ounry. Tking he reiprol, we shll mesure [ λ ( I ) / λ ( I )] using expors of Mexio nd Chin o he Unied Ses. While i would be preferble o use heir worldwide expors, our d for he US re more disggrege nd llows for finer mesuremen of unique produs sold by one ounry nd no noher. Speifilly, for 1989-2001 we shll use he 10-digi Hrmonized Sysem lssifiion of impors. To mesure he rio [ λ ( I ) / λ ( I )], we need onsisen omprison ounry. For his purpose, we shll use he worldwide expors from ll ounries o he US s he omprison, whih re lso verged over yers. Denoe his omprison ounry by, so h he se I U, I is he ol se of vrieies impored by he US in over ll yers, nd p i q i is he verge vlue of impors for produ i (summed over ll soure ounries nd verged ross yers). By ggreging ross ounries nd over ime, we obin onsisen omprison se of good I h does no iself vry over ime. 1 When ompring ounry o he ggrege ounry, i is immedie h he ommon 1 In onrs, Feensr nd Kee (2004) mesure expor vriey eh yer relive o he se of produs impored ino he US h yer, whih n led o inonsisen ross-yer omprisons.
9 se of goods expored is I I I = I, or simply he se of goods expored by ounry. Therefore, from (3) we hve h λ ( I ) = 1, nd: Vriey λ ( I ) λ ( I ) = i I i I p p i i q q i i. (4) Noie h he denominor on he righ of (4) is ol US impors, summed over ll produs nd ounries, bu using verge impor vlues over ime. The numeror equls he vlue of impors in produs h ounry sells o he U.S., gin summed over ll soure ounries nd verged over ime. This expression for he expor vriey of ounry, Vriey is herefore inerpreed s he shre of ol US impors from produs h re expored by ounry. Noe h his mesure depends on he se of expors by ounry,, bu no on is vlue of expors (exep insofr s hey ffe he vlue of worldwide expors). In he following seions we doumen expor vriey from Mexio nd Chin o he US, s mesured by (4) over 1990-2001. We lso disuss how riffs hve hnged for hose ounries, nd sisilly rele he hnge in expor vriey o he hnge in riffs. I 4. MEXICO'S EXPORT VARIETY: 1990 VERSUS 2001 We brek down he ggrege expors of Mexio o US ino seven mjor groups nd onsru he expor vriey indexes of hese seven indusries ording o (4). Tble 1 presens he indexes of hese indusries in 1990 nd 2001 o illusre he vriion ross indusries nd yers. In griulure, for exmple, Tble 1() shows h in 1990 Mexio expored 42 per en of ll he produ vrieies h he US impored (from ny ounry). Th shre inresed o 51 per en in 2001, for n nnul verge growh re of 1.8 per en. During he sme period, he
10 expor vriey from Mexio in he exiles nd grmens indusry inresed by 1.4 per en nnully, from overing 71 per en of ll vrieies impored by he US in 1990 o 83 per en in 2001. The highes growh re of vriey is in he eleronis indusry, where in 1990 only 40 per en of US impors were from produs expored by Mexio, nd in 2001 h shre is 66 per en. This represens n verge nnul growh re of 4.6 per en. On he oher hnd, he slowes growh of expor vriey is observed in he mhinery nd rnspor indusry, wih n nnul growh re of 1.3 per en. Th indusry long wih exiles nd grmens lredy hd high expor vriey in 1990, whih limied heir fuure growh in vriey. By onrs, expor vriey in he mining nd bsi mels indusry is mong he lowes mong ll non-griulure indusries in boh yers. Expors in h indusry overed 47 per en of he vrieies in US impors in 1990, nd inresed o 56 per en in 2001. In summry, over he ps dede Mexio hs expnded is expor vriey ross rnge of differen indusries. Averging over he indusries, 67 per en of US impor vrieies in 2001 re from produs whih Mexio expors, wheres h shre ws 52 per en in 1990. Could rde liberlizion explin he expnsions in expor vriey of Mexio over his smple period? Given h Mexio joined he Norh Ameri Free Trde Are (NAFTA) in 1994, we n ompre he verge vriey wihin eh indusry before nd fer 1994 o idenify he NAFTA effe on expor vriey. Figure 3 presens he verge vriey indexes pre nd pos-1994 in eh of he seven indusries. A he indusry level, he inreses rnge from 5.2 per en in he griulure indusry o 21.4 per en in he eleronis indusry. Overll, expor vriey of Mexio from inresed by 11.4 per en sine 1994. Th effe is sisilly signifin nd robus o indusry fixed effes, s we onfirm in ler seion.
11 Tble 1(b) presens he riff liberlizions in Mexio he indusry level in he pre nd pos-1994 ers wih respe o produs from he US. In 1990, he verge riff levied on US produs enering Mexio ws 12.1 per en, nd by 2001 h figure dropped o 1.1 per en. The mos drmi reduions n be found in he exiles nd grmens indusry, whose riffs dropped from 17.7 per en o 0.5 per en. Lrge riff dereses lso ourred in mhinery nd rnspor nd in eleronis. The indusry h hs he smlles reduion is griulure, whih sill hieved n reduion from 8.5 per en o 2.1 per en over he 11 yer period. Thus, Mexio wen hrough some very drmi delines in riffs wih respe o goods from he US. Tble 1() furher presens he riff liberlizions in he US he indusry level in he pre nd pos-1994 ers wih respe o Mexio s expors. In 1990, he verge riff imposed by he US on Mexio s produ ws 4.1 per en, nd in 2001, i ws 0.3 per en. In erms of bsolue hnges, he mos lrges fll in riff is in he exiles nd grmens indusry, where he verge riff dropped from 13 per en o 0.4 per en. In griulure nd eleronis, he fll in riffs is similr o he overll verge, nd in oher indusries he fll in riffs is less. A hllenge for our empiril work will be o explin he subsnil inrese in expor vriey over 1990-2001 using he relively smll (bu permnen) drop in US riffs under NAFTA. We reurn o his sk fer reviewing he expor vriey nd riffs for Chin. 5. CHINA S EXPORT VARIETY: 1990 VERSUS 2001 Tble 2() presens he expor vriey of Chin in 1990 nd 2001. The produ vrieies impored by he US h re lso expored by Chin rnge from 30 per en in griulurl indusry o 79 per en in he exiles nd grmen indusry in 1990, while in 2001, he rnge is beween 34 per en nd 88 per en. Over he sme period, he fses growh in expor produ vriey is in he mhinery nd rnspor equipmen indusry, wih verge nnul growh re of 7.3 per en.
12 Compred o Mexio, in 1990 Chin ws leding in erms of expor vriey in he exiles nd grmens indusry (wih 8.2 per en more expor vriey hn Mexio), nd wood nd pper produs (4.9 per en more expor vriey). On he oher hnd, Mexio ws hed in he mhinery nd rnspor equipmen indusry (37.5 per en more vriey), peroleum nd plsis (16.2 per en more), mining nd bsi mels (15.5 per en more), griulurl (12 per en more), nd eleronis (4.3 per en more). By 2001, he dvnge of Chin over Mexio in exiles nd grmens nd wood nd pper produs were redued o 5 per en nd 2 per en, respeively. Thus o he exen h NAFTA used n expnsions in Mexio s expor vriey, he mos drmi effes re in hese wo indusries. Despie he expnsion in Mexin expor vriey ross ll indusries, however, Chin ugh up in hose ses where Mexio led in 1990. In f, for he eleronis indusry, Chin s expor vriey exeeded h of Mexio by 2.5 per en in 2001, while he gps in he peroleum nd plsis, mining nd bsi mels, nd mhinery nd rnspor equipmen were redued o 2.5 per en, 1.3 per en nd 13.1 per en, respeively. Tble 2(b) presens he impor-weighed verge riffs he indusry level for Chin s expors o he US. I is eviden h while Mexio ws liberlizing is riffs due o NAFTA, Chin ws unilerlly reduing is riffs, oo. The verge Chinese riff on US produs dropped from 22.9 per en in 1992, he firs yer riff d is vilble, o 18.1 per en in 2001. The bigges reduions were in he mos proeed indusries, whih re exiles nd grmens nd mhinery nd rnspor equipmen. For he res of he indusries wih he exepion of griulure, he riff level is lose o 10 per en in 2001. The griulure indusry in Chin ws proeed by non-riff brriers (NTBs) suh s impor liensing nd quos, in ddiion o he high riffs. Figure 4 shows he exen of non-riff
13 brriers (i.e. quos nd ny oher impor resriions) minined by Chin in 1996. For rie nd wh, for exmple, 100 per en of he impors were subje o non-riff brriers. Under is enry o he WTO, Chin mde subsnil progress in reduing he NTBs in griulure, whih resuled in inreses in impor volume of goods in hose binding produs. Suh inreses pushed up he impor weighed verge riff in griulure from 14 per en o nerly 70 per en, s shown in Tble 2b. This does no refle n inrese in proeion, however, sine he unweighed verge riff in his indusry hs dropped from 48 per en o 27 per en (see Bhshli e l, 2004, for disussion). Tble 2() shows he impor weighed verge riffs of he US on Chin s expors he indusry level. Given h US does no hve ny rde greemen wih Chin, hese riffs refle he Mos Fvorie Nions (MFN) riffs. Overll, he verge riff on Chin s produ ws 5.8 per en in 1990, nd i dropped o 3.6 per en by 2001. Indusries h hve he lrges reduions re wood nd pper produs nd he eleronis indusry. In mos indusries, however, he perenge poin reduion in US riffs on Chin is no greer hn h for Mexio under NAFTA, nd someimes less. Averge US riffs on Chin in griulure nd in peroleum nd plsi produs hve inresed very slighly. 2 6. REGRESSION RESULTS To deermine he effes of rde liberlizion on expor vriey, we fous on he se of Mexio. Tble 3 presens he regression resuls. We pool observions ross indusries nd yers, using he full smple from 1974-2001 for he iniil regressions. 3 There re seven 2 The inrese in verge riffs in hose indusries mos likely refles shifing impor bundle in he US owrds produs wih slighly higher riffs. 3 For 1974-1988 we onsru expor vriey using he 7-digi Triff Shedule of he Unied Ses, Annoed. Beuse h lssifiion differs from he Hrmonized Sysem used fer 1989, he expor vriey indexes re inonsisen beween 1988 o 1989. So we re-sle he erlier indexes so h for eh indusry nd eh ounry,
14 indusries nd 28 yers, whih forms blned smple of 196 observions. In olumn (1), we regress he log of indusry expor vriey of he indusries on he NAFTA indior vrible, whih se o one for 1994 nd ler yers, nd zero oherwise. Conrolling for indusry fixed effes, he NAFTA indior is sisilly signifin, whih indies n inrese in expor produ vriey. The esimed oeffiien is 0.20, whih implies 20 per en inrese in expor vriey due o NAFTA. We sudy he pril effes of he US riff reduion on Mexio s expor vriey in olumn (2). The NAFTA indior vrible my be piking up oher fors h hnge over ime monoonilly, whih will bis our esimes. Thus, in ddiion o he NAFTA indior, we inrodue he log vlue of one plus he US riffs. Conrolling for indusry fixed effes, he riff erm is negive nd sisilly signifin. The oeffiien on he US riff in olumn (2) is inerpreed s semi-elsiiy: eh one perenge poin reduion in US riffs inreses expor vriey from Mexio by wo per en. This esime of he semi-elsiiy indies h riff us in he US re imporn in inresing expor produ vriey of Mexio, bu hey nno explin he observed inrese in vriey fer NAFTA. Noie h he NAFTA indior in olumn (2) is sill esimed 0.14, showing h he h he US riff us only explin 6 perenge poins of he expnsion in expor vriey from Mexio, wih he remining 14 per en ribued o some oher NAFTA effe. Anoher wy o rrive his onlusion is o muliply he semi-elsiiy of wo by he verge drop in US riffs, whih is 4.1 0.3 = 3.8 per en from Tble 1(), rriving predied inrese in verge expor vriey of 7.6 per en. This is only bou one-hlf of he ol inrese in expor vriey repored in Tble 1(), of 66.7 52.4 = 14.3 per en. expor vriey in 1988 equls h in 1989. In ddiion, we inlude n indior vrible for 1989 in he regressions, o furher onrol for he hnge in lssifiion sysems.
15 The resuls in olumns (1) nd (2) my be bised if here re omied vribles h re orreled wih he NAFTA liberlizion nd wih he expnsion of expor vriey. One suh vrible is he negive mrke ompeiion effe due o he expnsion of oher ounries expor vriey. To he exen h he reduion of he US MFN riff uses n expnsion in Chin s expor vriey, we expe Mexio s expors o be rowded ou nd is expor vriey o derese. Given h he rde liberlizion of Chin oinided wih he Mexio s liberlizion, omiing he ompeiion from Chin my use downwrd bis on he esimed oeffiien of he US riffs on Mexio. We onrol for he mrke ompeiion effe from Chin by inluding he indusry expor vriey of Chin s n explnory vrible. Column (3) inludes he expor vriey of Chin wihin he NAFTA indior speifiion, long wih he US riff reduion. I is ler h inluding Chin s expor vriey resuls is subsnil lowering of he NAFTA imp, o 4% in olumn (3), whih is insignifinly differen from zero. However, he resuls in olumn (3) ould be bised iself due o he endogeneiy of Chinese expor vriey. As menioned before, inluding Chin s expor vriey my pik up he mrke ompeiion effe, whih leds o negive effe on he expor vriey of Mexio. Bu he expnsion of Chin s expor vriey my lso be driven by indusry-speifi ehnologil progress or US demnd shoks h re ommon beween he wo ounries wihin indusry nd yer. This would hve posiive effe on he expor vriey of Mexio. While indusry-speifi ehnologil progress is unobservble, he expnsion of Chinese expor vriey due o Chin s riff reduions n be used o pure he mrke ompeiion effe. We sudy his hypohesis by using Chin s indusry riffs s n insrumenl vrible for Chinese expor vriey. When he speifiion in olumn (3) is run using his insrumen, he
16 resuls (no repored in Tble 3) re quliively similr: he oeffiien on Chinese expor vriey is sill posiive, so h we re no piking up mrke ompeiion effe. This ppers o be due o he long ime spn of he smple, for 1974 o 2001. In olumns (5) nd (6) we use shorer ime period, from 1990 o 2001, whih he fous of his sudy. We run sysem of wo equions, one for he expor vriey of Mexio nd noher for he expor vriey of Chin. In boh equions, he se up is idenil o h of olumn (4), whih inludes he US riff, he expor vriey of he oher ounry, nd indusry speifi effes whih re reed s ommon ross he equions. Expor vriey of he oher ounry is endogenous, nd s n insrumenl vrible we use he US riff on h ounry. Column (5) shows he equion for Mexio s indusry expor vriey, nd (6) shows he equion for Chin s indusry expor vriey. We now find h he mrke ompeiion effe of Chinese produs on Mexio expors re negive nd sisilly signifin. Every one per en inrese in he expor vriey of Chin redues he expor vriey of Mexio by one-hlf of one per en, in olumn (5). On he oher hnd, n expnsion of Mexio s vriey does no hve signifin imp on Chin s expor vriey, in olumn (6). Conrolling for he expnsion in Chinese produs due o riff reduions in Chin nd he US, he mrginl effe of he US riff liberlizion in olumn (5) is lrger hn is ws before: eh one perenge poin reduion in riffs now inreses Mexin expor vriey by 4.5 per en. This demonsres he subsnil imp of riff liberlizion on produ vriey of he exporing ounry. To see how muh of Mexio s expor vriey inrese we re now explining, we go hrough similr lulion s before. We muliply he semi-elsiiy of 4.5 by he verge drop in US riffs, whih is 3.8 per en from Tble 1(), rriving predied inrese in verge expor vriey of 17.2 per en. This is slighly lrger hn he ol inrese in expor vriey
17 repored in Tble 1(), of 14.3 per en, so h he US riff u fully explins he verge inrese in expor vriey from Mexio. However, regression (5) lso predis fll in Mexio s expor vriey due o he ompeiive imp from Chin of 0.5 21.2 = 10.6 per en. 4 So in ol, he regression under-predis he verge inrese in Mexio s expor vriey. Performing he sme lulion on speifi indusries, regression (5) over-predis he inrese in expor vriey in exiles nd pprel from Mexio, nd under-predis mos oher indusries. Evidenly, exiles nd pprel is n oulier wih very subsnil drop in US riffs on Mexio bu modes inrese in expor vriey. The f h his indusry performs differenly from he ohers lso indies h he regression should be run seprely ross indusries. Th is beyond he sope of he presen pper due o lk of observions, bu would be possible in pnel dse wih more ounries, whih is n imporn direion for furher reserh (Debere nd Mosshri, 2005). 7. CONCLUSIONS The ps dede hs winessed signifin inrese in Mexio s expor vriey in ll indusries, espeilly sine NAFTA wen ino effe. Overll, 67 per en of US impors in 2001 re from produs whih Mexio expored, wheres his shre ws 52 per en in 1990. Over he sme period of ime, Chin lso experiened rpid expnsion in expor vriey, nd in erin indusries, Chin now exeeds Mexio in erms of expored vrieies. In his pper, we sudy he effes of US riff reduions on expor vriey. Our empiril resuls indie h riff liberlizion is imporn in expnding expor vriey. In priulr, here is sisil evidene linking US riff liberlizion due o NAFTA o inresed expor vriey from Mexio. Th effe is robus o he mrke ompeiion effe of Chinese expors, nd 4 This moun is he elsiiy of Mexio s expor vriey wih respe o Chin s expor vriey, imes he verge inrese in Chin s expor vriey, whih is 63.3 42.1 = 21.2, from Tble 2().
18 in f, he semi-elsiiy beween riff us nd expor vriey is higher when he ompeiion from Chinese expors is ken ino oun. While he si gins from rde hve been widely sudied nd doumened o be relively smll, he dynmi gins due o he expnsion of expor vriey my well be more imporn. Brod nd Weinsein (2004) doumen h he expnsion of impor vrieies in he Unied Ses hve hd signifin imp on lowering he rue impor prie index, nd herefore on rising US welfre. Similrly, Feensr nd Kee (2006) rgue h he growh of expor vrieies benefis ggrege produiviy in he exporing ounry. This pper shows how he expnsion in he vriey of rded goods is linked o riff reduions, hereby onribuing o shor erm nd long erm gins.
19 REFERENCES Bernd, E. R., M. K. Kyle, nd D. Ling (2002) The Long Shdow of Pen Expirion: Generi Enry nd Rx o OTC Swihes, in Rober C. Feensr nd Mhew Shpiro, eds., Snner D nd Prie Indexes, NBER Sudy in Inome nd Welh (Chigo: Univ. of Chigo), 229-267. Bhshli, D., S. Li nd W. Mrin (2004) Chin nd he WTO: Aession, Poliy Reform, nd Povery Reduion Sregies (The World Bnk nd Oxford Universiy Press). Brod, C. nd D. Weinsein (2005) Globlizion nd he Gins from Vriey, Qurerly Journl of Eonomis, forhoming. Brod, C., J. Greenfield nd D. Weinsein (2006) From Groundnus o Globlizion: A Sruurl Esime of Trde nd Growh, Universiy of Chigo nd Columbi Universiy. Debere, Peer nd Shlh Mosshri (2005) Do Triffs Mer for he Exensive Mrgin of Expors? An Empiril Anlysis, Universiy of Texs, Ausin. Dollr, D., nd A. Kry (2001) Trde, Growh nd Povery, Poliy Reserh Working Pper #2615 (The World Bnk). Feensr, R. C. (1994) New Produ Vrieies nd he Mesuremen of Inernionl Pries, Amerin Eonomi Review, 84(A1), Mrh, 157-177. Feensr, R. C., D. Mdni, T. Yng nd C. Ling (1999) Tesing Endogenous Growh in Souh Kore nd Tiwn, Journl of Developmen Eonomis, 60, 317-341. Feensr, R. C. nd H. L. Kee (2004) On he Mesuremen of Produ Vriey in Trde, Amerin Eonomi Review, 94(2), 145-149. Feensr, R. C. nd H. L. Kee (2006) Expor Vriey nd Counry Produiviy: Esiming he Monopolisi Compeiion Model wih Endogenous Produiviy, World Bnk Poliy Reserh Group nd Universiy of Cliforni, Dvis. Frnkel, J. A. nd D. Romer (1999) Does Trde Cuse Growh,? Amerin Eonomi Review, 89(3), 379-399. Funke, M. nd R. Ruhwedel (2001) Produ Vriey nd Eonomi Growh: Empiril Evidene from he OECD Counries, IMF Sff Ppers, 48(2), 225-242. Grilihes, Z. nd I. M. Cokburn (1994) Generis nd New Goods in Phrmeuil Pries Indexes, Amerin Eonomi Review, 84(A5), Deember, 1213-1232. Grossmn, G. M. nd E. Helpmn (1991) Innovion nd Growh in he Globl Eonomy, Cmbridge: MIT Press.
20 Husmn, J. A. (1997) Vluion of New Goods under Perfe nd Imperfe Compeiion, in Timohy F. Bresnhn nd Rober J. Gordon, eds., The Eonomis of New Goods, NBER Sudies in Inome nd Welh, no. 58 (Chigo: Universiy of Chigo, 209-237). Husmn, J. A. (1999) Cellulr Telephone, New Produs, nd he CPI, Journl of Business nd Eonomi Sisis, 17(A2), April, 188-94. Hiks, J. R. (1940) The Vluion of Soil Inome, Eonomi, 7, 105-24. Hummels, D. nd P. Klenow (2005) The Vriey nd Quliy of Nion s Trde, Amerin Eonomi Review, 95(3), June, 704-723. Kehoe, T. J. nd K. J. Ruhl (2002) How Imporn is he New Goods Mrgin in Inernionl Trde? (Universiy of Minneso nd Federl Reserve Bnk of Minnepolis). Rodriguez, F. nd D. Rodrik (2000) Trde Poliy nd Eonomi Growh: A Skepi s Guide o he Cross-Nionl Evidene, in Ben S. Bernnke nd Kenneh Rogoff, eds., NBER Mroeonomis Annul 2000, 261-325. Romer, P. (1990) Endogenous Tehnil Chnge, Journl of Poliil Eonomy, 98(5, Pr 2), S71-S102.
21 Averge Tble 1: Mexio s Trde wih he U.S., 1990-2001 Agriulure () Mexio s Expor Vriey o he U.S. Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 52.4 41.5 71.2 47.3 55.4 46.6 65.6 39.5 2001 66.7 50.9 82.6 63.2 72.7 56.4 75.8 65.6 Growh re 2.2 1.9 1.4 2.6 2.5 1.7 1.3 4.6 Overll Indusry Agriulure (b) Mexio s Triffs on Impors from he U.S. Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 12.1 8.5 17.7 10.9 9.9 9.7 14.0 13.9 2001 1.1 2.1 0.5 0.6 1.5 1.3 1.1 0.6 Overll Indusry Agriulure () US Triffs on Impors from Mexio Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 4.1 4.4 13.0 2.2 0.6 2.1 2.5 4.1 2001 0.3 0.8 0.4 0.0 0.1 0.7 0.1 0.1 Noes: 1. All figures re in per en. 2. The growh re in () is ompued s [ln(vriey 2001 ) ln (Vriey 1990 )] 100. Soures: Expor vriey in () nd US riffs in () re ompued from he US impor d desribed in Feensr, Romlis nd Sho (2002). Triffs in (b) re ompued from he WITS dbse he World Bnk.
22 Averge Tble 2: Chin s Trde wih he U.S., 1990-2001 Agriulure () Chin s Expor Vriey o he U.S. Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 42.1 29.6 79.4 52.2 39.2 31.1 28.1 35.2 2001 63.3 34 87.6 65.2 70.3 55.1 62.7 68.1 Growh re 3.7 1.3 0.9 2.0 5.3 5.2 7.3 6.0 Averge Agriulure (b) Chin s Triffs on Impors from he U.S. Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 22.9 14.0 52.6 15.8 17.8 17.0 33.7 18.3 2001 18.1 69.8 17.6 10.6 11.3 9.0 11.4 10.7 Averge Agriulure () US Triffs on Impors from Chin Texiles & Grmens Wood & Pper Peroleum & Plsis Mining & Mels Mhinery &Trnspor Eleronis 1990 5.8 1.4 13.0 6.6 2.7 6.4 5.1 5.1 2001 3.6 1.7 10.9 1.3 3.1 4.1 3.0 1.2 Noes: 1. All figures re in peren. 2. Chinese riffs lised in 1990 re for 1992. Oher noes nd soures from Tble 1 lso pply.
23 Tble 3: Dependen Vrible Log of Indusry Expor Vriey (1) (2) (3) (4) (5) (6) OLS OLS OLS IV 3SLS 3SLS NAFTA indior 0.198** 0.143** 0.043-0.006 (0.036) (0.044) (0.033) (0.034) ln(1+mexio riff) -2.049* -2.109** -2.139** -4.535** (0.997) (0.540) (0.460) (0.932) Chin's expor vriey 0.316** 0.474** -0.508** (0.046) (0.042) (0.108) ln(1+chin riff) -2.974* (0.897) Mexio's expor vriey 0.116 (0.142) Consn 3.520** 3.630** 2.669** 2.782** 5.70** 3.11** (0.044) (0.069) (0.137) (0.148) (0.391) (0.566) Indusry fixed effes Yes Yes Yes Yes Yes Yes No. of observions 196 196 196 196 84 84 R-squred 0.50 0.51 0.78 0.71 Chi-squre 131.2 241.4 Noes: 1. Whie-robus sndrd errors re in prenheses. * signifin 5%; ** signifin 1% level. 2. Dependen vribles re he log of Mexio's expor vriey in olumns (1)-(5), nd he log of Chin's expor vriey in olumn (6). Columns (4), (5) nd (6) re esimed wih insrumens for expor vriey, whih onsis of h ounry's riffs. Columns (5) nd (6) hve ommon indusry fixed effes. 3. Smples over 1974 o 2001 in olumns (1) o (4), nd 1990 o 2001 in olumns (5) nd (6). There re seven indusries inluded, s shown in Tbles 1 nd 2.
24 x 2 D B C A x 1 Figure 1: Inpu Vrieies x 2 B C A x 1 Figure 2: Oupu Vrieies
25 Pre NAFTA Pos NAFTA Expor Vriey (%) 90 80 70 60 50 40 30 20 Agriulure Eleronis Wood & Pper Texiles & Grmens Mining & Bsi Mels Peroleum & Plsis Mhinery & Trnspors Figure 3: NAFTA nd Expor Vriey Soure: Auhor s lulions 100 Peren Coverge 80 60 40 20 0 Trnspor Indusries Ligh Mnufures Wering Apprel Texiles Nurl Resoures Oher Food Produs Me nd Milk Livesok Nongrin Crops Corse Grins Whe Rie Bsi Hevy Mnuures Mhinery nd Equipmen Tol Servies Figure 4: Non-riff Mesures Affeing Chin s Impors Soure: World Bnk, Chin 2020: Chin Engged.