Auto Parts Trade: Elasticity of Substitution in a Third Market

Transcription

1 (Preliinary draft) Paper prepared for the Global Eonoy International Conferene Marh 17-18, 2012, University of Washington, Seattle, U.S. Auto Parts Trade: Elastiity of Substitution in a Third Market By Iwao Tanaka, Toyo Gakuen University, Tokyo, Japan 1 Abstrat: This study attepts to eaine whether an autoobile part eported fro Japan is a kind of speial and hardly replaed by the sae part eported fro other ountries by estiating the elastiity of substitution between Japan and opeting ountries in a third arket. The reent literature on foreign outsouring and inoplete ontrat suggests that international transation of interediate inputs take plae through one or any obination of the following three fors: intra-fir trade, international outsouring with a ontrat, or international outsouring at ar s length. I fous on the last two options in this paper, that is, a how-to-buy deision proble. In general, a aker would hoose outsouring with a ontrat over arket purhases, when the desired part is speifi enough to be assebled together with partiular parts and oponents for a ar. However, when there are other ountries who an supply the sae part at relatively lower ost, and a ontrat agreeent with a supplier breaks down, a aker would outsoure the part at ar s length. Assuing the Arington s hypothesis on the prodution loation, one an postulate that lower elastiity of substitution for a part fro a partiular ountry should ean a higher level of uniqueness of that ountry s part. I set up a regression odel based on the two-level onstant-elastiity-of-substitution (CES) prodution funtion and estiate it to eaine how unique Japanese auto inputs are in a third arket. It appears that soe auto parts eported fro Japan are less substitutable, eplaining the teporal disruption of a global supply hain funtioning for a otor vehile industry, even ausing a deline of U.S. autoobile prodution right after the earthquake on Marh 11th., in JEL Classifiation Nubers: F12, F14, F23, L23, L24 Keyword: trade in interediate inputs, elastiity of substitution, vertial network, ontrating environent, U.S. auto parts iports 1 Contat address is , Hongo, Bunkyo-ku, Tokyo Japan [ ], and (Phone). E-ail address is iwao.tanaka@tyg.jp. 1

2 1. Introdution International trade in interediate inputs is growing very fast. It aounts for about 40 perent of worldwide trade. A large part of the growth of suh trade is due to an inrease of international outsouring and vertial foreign diret investent (FDI), whih indues intra-fir trade as a result of etending prodution proesses aross any ountries. Intra-fir trade now aounts for about one third of worldwide trade. One of the eonoi ipats of the earthquake hitting the northern part of Japan on Marh 11th., 2011, was to reate a onsiderable daage on parts fatories of sall-and-ediu sized firs of this region. They produed parts and oponents of otor vehiles and various eletroni produts. Sine they played an essential role as a global part supplier, the global supply hain was fored to be teporarily disrupted after the natural disaster 2. The Ministry of Eonoy, Trade and Industry [METI], Japan, reports that due to this disruption, U.S. auto and auto-parts prodution leveled down by 8.9% in April 2011 (over previous onth) (METI (2011)). They argue that in partiular, any fatories of a iro ontroller unit 3, one of the essential eletri parts for otor vehiles, as well as for eletroni produts suh as a obile phone and reote ontroller of television set, were loated in the region and severely daaged 4, naturally leading to the disruption of their part supply and redution of auto prodution in the U.S. 5 This inident otivated y urrent researh asking whether and why suh inputs eported fro Japan are less substitutable in a third ountry. The literature on the fir s organization and inoplete ontrat provides an iportant idea about fir s ake-and-buy (and how-to-buy) deision aking in onnetion with the input trade. The theoretial arguent has been developed on the fators for whih vertial integration is preferred to international outsouring (Grossan and Helpan (2003), Spener (2005), Helpan (2006b), for eaple). For the eplanation, ontrating environent of a ountry and ontratible ativities in an input prodution are partiularly foused on reently (Aeoglu, Antras, and Helpan (2007), Nunn (2007), and Nunn and Trefler (2008)). Another foal point is about the odel struture: tehnologial opleentarities. 2 Disruption of the global supply hain had been already observed before the earthquake happened as one of the fators of the great trade ollapse, and as one of the ajor ipats of global finanial risis and great reession after the Lehan Brothers inident of 2008, aording to Feenstra (2011). 3 On the HS 9-digit lassifiation, iro ontroller units are under the line of , that is, a part of the 6-digit produt group (eletroni integrated iruits et.). 4 It is Ibaraki Prefeture. 5 The METI also reports that the world share of this part by Japanese firs represents about 30%, but that the inventory ratio is very low as a otor vehile prodution standard: only for 17 days. 2

3 Aeoglu, Antras, and Helpan (2007) introdued a odel in whih one an analyze the ipat of ontratual iproveent on the elastiity of input substitution depending on the degree of tehnologial opleentarity. Suh ipat is higher when the degree of tehnologial opleentarity aong inputs is greater. It is also eplained by Helpan (2006a) that industries with low elastiities are ore sensitive to the ontrat inopleteness. Based on those arguents, the urrent paper fouses on the substitutability of the sae input fro different ountries with ontrating enforeability taken aount for. As for epirial studies, only a few researhes have been onduted. Head, Ries, and Spener (2004) and studies by Nunn and Trefler are the ost representative. As Helpan suggests, the diret testing of the hypothesis fro the theory is diffiult beause of the data availability: we need transation data lassified for international outsouring with and without a ontrat as well as FDI on a fir level 6. Thus, industry data are ore often used to investigate whether the resulting trade pattern reflets suh fators as ontrating environent. Case in point, the share of U.S. intra-fir trade is investigated by Nunn and Trefler (2008). Using the Census data, fro whih iports and eports by related-parties are available 7, they find that for industries with high level of headquarter intensity, intra-fir iports are positively related to ontrating environent 8. I set up a odel in whih an iporter s prodution funtion depends on iported interediate inputs that are assued to be iperfetly substitute by the plae of prodution (Arington s assuption). With the assuption of strong separability aong inputs, I use the two-level onstant-elastiity-of-substitution (CES) prodution funtion introdued by Sato (1967). It is also assued that an input fro eah ountry is to soe etent haraterized by the degree of ontrating environent of that ountry. An estiation using U.S. iport data shows that for soe parts, in fat, the iro-ontroller-inluded produt ategory entioned above represents lower elastiity of substitution and strong ontrat inopleteness effet. Thus, it appears that this kind of part eported fro Japan is rather unique, opared to the sae part fro other ountries probably beause of its tehnial speifiity. 6 In addition, the inopleteness of a ontrat and the portion of ontratible ativities for input prodution are theoretially of iportane, for whih aquiring data ight be even ore diffiult in reality. 7 Aording to U.S. Census Bureau News by the U.S. Departent of Coere, in the U.S. the share of trade by related-party aounted for 40.8% of the total goods trade in Sine iproveent of the ontrating environent is ore likely to inrease outsouring with a ontrat, their result an be interpreted as the ase where-the surprise effet (slightly) doinates the standard effet (Nunn and Trefler (2008), p.75). 3

4 2. Eporter substitution in a third arket: autoobile parts and oponents To eaine the degree of substitutability for Japanese parts eports to a third arket, I first diretly estiate an elastiity of substitution for autoobile part eports between Japan and opeting ountries in U.S., China, and world arkets. The produt ategories used here are the ones generally onsidered to be used as parts and oponents for otor vehile prodution, whih are indiated under the Haronized Coodity Desription and Coding Syste (HS) 85 and 87 ategories of the United Nations, Coodity Trade Database [UN COMTRADE]. I seleted 18 6-digit auto parts. These HS ode nubers and their produt naes are listed in Appendi Table. The eporter elastiity of substitution ( ) is defined as follows. d ln J, (1) w d ln J w where J represents Japanese eport quantity of a part to a third ountry, opeting ountry s eport quantity of the sae part to the sae ountry, J is the w and are the pries of the part eported fro Japan and fro the opeting ountry, respetively. The part is assued to be iperfetly substituted. The siple way to obtain the elastiity of substitution is to regress the nuerator on the denoinator of equation (1). That is, ln J w ln J u w. (2) An error ter u is added to desribe an estiation odel in equation (2). The estiate of the oeffiient on the relative prie ter shows the elastiity estiated ( ). The sign of is defined as negative. It eans that for a part with large elastiity, if the prie of the part eported by a Japanese supplier is raised, its eport is ore likely replaed by the eport of the sae part by a supplier in the other ountry. If the elastiity is sall, however, the prie rise of the part is not likely to lead to the replaeent with the opetitor. The COMTRADE database provides us with detailed trade data lassified by-ountry-by-oodity iport and eport values and quantities. I olleted eport data of the 18 parts for 28 ountries and area, whih are designated to the U.S., China, and the world. A unit prie of eport is alulated siply by dividing the eport value by the eport quantity (nubers or weights (kilogra)). Japanese variable is taken as a nueraire, and so that, its relative ratio to eah one of the 27 opeting ountries is w 4

5 only alulated. Table 1 reports the estiation results 9. The goodness of fit is generally better for the U.S. odel than for the China s odel. Thus I look at the results of the U.S. odel. Conerning the agnitude of elastiities, it is found that they vary and ostly range fro 1.76 (HS , starter otors et.) to nearly 4 (HS , steering wheels et.) in ters of absolute value. This suggests that tehnologial harateristis of a part and other fators give rise to the different degree of eport opetition. As for the produt ategory, HS , the produt ategory inluding iro ontroller units, the elastiity is estiated to be -1.99, whih is the one of the lowest elastiity group aong 18 parts. When this part is eported fro Japan, it is very unlikely to be replaed by the sae part eported fro other ountries. The arguent so far was ade fro an eporter s point of view. Sine, to large etent, parts trade refleted a final-good aker s iport deision rather than the eporters, I swith the viewpoint fro that of eporters to that of iporters in the net setion. This is aied at odeling input proureent options with the ontrat inopleteness as suggested by the literature. 3. Vertial networks and elastiity of substitution for auto parts In the growing international input trade, vertial prodution networks play a signifiant role in deterining the fir s deision-aking on a ake-and-buy and how-to-buy: this deision inludes a fir s organizational hoie proble, suh as outsouring versus vertial integration, or in partiular, outsouring to an insider versus outsouring to an outsider. There are also other options: a aker ould produe parts in-house rather than to outsoure, or it ould also hoose outsouring to doesti suppliers, or vertially integrate with doesti suppliers. 3.1 Outsouring, relationship-speifi investent, and inoplete ontrat Aong those several possible deision patterns, I fous on the insider-outsider odel well eplained in Head, Ries, and Spener (2004). The odel is the one where a aker deides whether to outsoure a part to a (foreign) supplier who oits a relationship-speifi investent (RSI) based on an inoplete ontrat (insider), or to outsoure fro an (foreign) independent supplier at ar s length (outsider). In this 9 The nuber of observations varies fro 13 to 26 for the U.S. and China s odel, and 14 to 27 for the World odel, depending on how any of the 27 opeting ountries atually eported. 5

6 odel the vertial integration through foreign diret investent (FDI) is not onsidered as a proureent option. The fat that it is not so often seen when Japanese parts suppliers are integrated with U.S. auto akers by FDI provides a good reason that vertial integration is not a very iportant option as far as iports fro Japan are onerned. The relationship-speifi investent (RSI) plays a entral role. It is viewed that suh investent is required for part-suppliers to anufature high-quality parts or parts used only for speifi odels. 10 A aker prefers to outsoure a part that would be epensive if it were ade by theselves, to an independent supplier so as to redue prodution ost. The iportant thing is that the RSI ust be governed by a ontrat. It should be noted, however, that the ontrat is inoplete sine whether the onditions are satisfied an t be iediately verified by either party (at the first stage of the gae). Therefore, eistene and availability of a third party for verifiation or reliable legal systes is partiularly iportant in deterining whether to outsoure with a ontrat. The ontratual part souring also suggests that the prie of a part an be negotiated and renegotiated at the seond stage (Feenstra and Spener (2006)), and one it is set, the sae prie ould last until the terination of the ontrat date oes. The feature of y odel relies on the two basi harateristis of the odel with the RSI and inoplete ontrat. First, parts are lassified and ordered aording to the effiay of RSI, whih eans an inreent of rents generated by RSI on the supplier s side but obtained by a aker ontributing to a derease in its arginal ost of produing the part. The effiay is speifi to a part, and it would also be affeted by other fators suh as business networks, like Japanese keiretsu, through whih inforation ay be easily ehanged, ontrat enforeability beause of the nature of the ontrat inopleteness, and so on. Seond, parts that require ore RSI are likely to be used for a partiular good suh as a luury passenger ar, and to be used together with other parts and oponents to be assebled, like a ylinder and an engine. A part required of ore of RSI ust be neessary for other partiular parts to be used together and assebled as a ar. It follows that the degree of opleentarity between those parts should be very high (for eaple, the elastiity between a ylinder and an engine), and that the degree of substitutability for the part between suppliers with different ountries of origin should be lower (for eaple, the elastiity between a ylinder iported fro Malaysia and a ylinder iported fro Japan an be different 10 Asanua (1989) tered the relationship-speifi investent for the investent in skills that require...the supplier to respond effiiently to the speifi needs of the ore fir. Prevalene of this kind of investent and subontrating is a well-known business syste aong keiretsu ebers espeially during the high growth period of Japanese eonoy, whih is detailed by Yaawaki (2007). 6

7 due to the tehnial speifiity and its use in atual prodution in response to the requireent of the RSI by a aker). The following proposition suarizes the arguents. Proposition When the relationship-speifi investent is iportant for a partiular part prodution, (1) the part is likely to be produed by a supplier loated in a ountry with a relatively better ontrating environent, (2) if it is iported by a third ountry, the elastiity of substitution for the part between ountries should be low (intra-lass elastiity of substitution), and (3) sine the part is tehnially speifi to a partiular ar, it ust be used with other speifi inputs to asseble a ar, and thus, the elastiity of substitution between parts should also be low (inter-lass elastiity of substitution). In the net sub-setion, I disuss an analytial fraework by whih the above proposition an be eained. 3.2 Analytial fraework Let us assue that a representative U.S. auto aker produes an autoobile by using interediate inputs only. All inputs are assued to be iported. They are also distinguished by the plae of prodution, as suggested by Arington, and thus, they should be iperfet substitutes for eah other. With the strong separability assuption aong the parts, this prodution tehnology an be represented by the following two-level onstant-elastiity-of-substitution (CES) prodution funtion. where M 1 q M M 1, (3) 1 r. (4) 1 In the prodution funtion of the final good (equation (3)), q is the auto output, whih depends on kinds of iported parts, M. ' s are distribution paraeters. The part prodution funtion ( M ) is given by equation (4), whih depends on, the part iported fro ountry. To the etent that the ontrating enforeability is speifi to eah ountry, I assue that an iport volue fro a 7

8 ountry be augented by the degree of ontrating environent of that ountry. This effiieny of RSI is aptured by the following for: r, that is, an effiieny unit of iported part fro ountry, where r represents the degree of ontrating environent of ountry. are the distribution paraeter. and are the substitution paraeters. The usual paraeter onditions are also assued: 0, 0, 0, 1, L K 1 1, and 0 1 We all the inter-lass elastiity of substitution, while of substitution., 1, the intra-lass elastiity This funtional for is first introdued by Sato (1967), and proved to preserve the onstany feature of the elastiity. The advantage of this funtion lies in the separability of optiization probles. At first stage a aker deides how any units of a partiular input it uses solely based on the relative input prie to the other input, and at the seond stage, it deides how any units of the first input it iports fro a partiular ountry elusively within the first input 11. In our fraework, with the total auto prodution held onstant, an auto aker deides whether to outsoure with a ontrat does not depend on the deision of how any units of the part is neessary. We first onsider the optial ondition at the seond stage. The least ost ondition ust be satisfied for eah pair of Japan and a opeting ountry. The ratio of arginal produt of a part fro ountry relative to that of the sae part iported fro Japan ust be equated to the ratio of the unit iport pries, ountry over Japan. q q w. J wj Based on equation (4), we have the following ondition. 1 w w J J r r J J 1. (5) Thus, for a part, we have, J J r r J 1 w w J. (6) Defining X, J B, J r R r J, and w w J, rewrite the equation. 11 Sato (1967) desribes for this advantage the hoie of the ost-iniizing fator obination is effetively separated into two stages. 8

9 X B R 1. (7) Taking a natural logarith of equation (7), a redued for of the estiation is obtained as ln X a0 a1 ln R a2 ln u, (8) where a0 ln B, a1 1, and a. 2 The intra-lass elastiity of substitution ( ) an be obtained as an estiate of the oeffiient on ln in equation (8). Net, the optiality ondition at the first stage is derived as follows. The least ost ondition ust hold for eah pair of the parts, and n. where Defining p and q q p n, pn p are the pries of parts and n, respetively. n Based on equation (3), we obtain M n and M M p P n p 1 M n n pn. (9) M p estiation equation beoes, ln M b0 b1 ln Pu, (10) where, and taking a natural logarith, then, a redued for n b ln, and b 0 1. The inter-lass elastiity of substitution ( ) an be obtained as an estiate of the oeffiient on ln P in equation (10). 3.3 Estiation and Data The epirial appliation using the two-level CES prodution funtion an be seen in the literature. Saito (2004) is one of the studies, in whih she estiated both inter- and intra- group elastiities of substitution for two-digit industries to argue on the plausibility of the use of bilateral trade data for testing the Arington hypothesis. Her estiation of the intra-group elastiity onsists of eah obination of bilateral trade and prie ratios, that is, they are not liited to the ratio of iport volue or prie relative to Japan, different fro this study. U.S. parts iport data are obtained again fro the United Nations, COMTRADE database. I olleted iport values and quantities for the sae 18 HS-6-digit produts as the previous setion, whih are onsidered to be used for otor vehiles prodution. 9

10 The iport data is bilateral with all ountries fro whih the U.S. iported a partiular part in 2010, and iports fro Japan are reorded for all of the 18 parts. The data on ontrating environent does not eist as suh, but very often in the literature, the rule of law inde by Kaufann, Kraay, and Mastruzzi (2009) is used as a proy. Following the Nunn and Trefler (2008), I olleted the indies for all available ountries with the inde ranging fro -2.5 to +2.5, and odified the so that the inde lies between 0 and 1 (by adding 2.5 and divide it by 5). This data is for The ratios in estiation equations (8) are defined as the opeting ountry over Japan. The estiate of a 1 in equation (8) is of iportane in the sense that it iplies the effet of the ontrating environent on the part iport. We an argue on this point based on the Proposition (1) desribed in sub-setion 3.1. The estiate of a 2 in equation (8), on the other hand, shows the agnitude of the intra-lass elastiity of substitution between Japan and the opeting ountry for a partiular part, as suarized in Proposition (2). Finally, the estiate of b 1 provides the inter-lass elastiity of substitution for eah pair of two parts, as suggested in Proposition (3). 3.4 Estiation results The estiation results are reported in Table 2. Unlike the previous estiation, the nuber of parts had to be redued to five fro 18. This is priarily beause the unit prie alulated fro the U.N. database does not provide enough variations, and the virtually onstant unit prie aross ountries sees to oe fro the trade-volue (or value) estiation ethod onduted by the U.N. As a result, I estiated the odel for the five parts only, whose unit prie has an enough variation 12. Without the ontrat variable, the intra-lass elastiity estiates are negative and signifiant for the three out of five parts, ranging fro (HS , eletroni integrated iruits et.) to (HS , safety seat belts et.). The elastiity estiated for the integrated iruits, whih also inludes iro-ontrollers, shows the lowest agnitude. The low elastiity of substitution for the part learly indiates that the part iported fro Japan is less substitutable by the sae part fro other ountries in the U.S. When the ontrating environent variable is inluded in estiation, the agnitude of the elastiity beoes slightly even saller (-1.43) with a statistially and signifiantly positive estiate of the ontrating variable. If the opeting ountry is haraterized by a worse ontrating environent relative to Japan, the part iport fro Japan would inrease. 12 Those five parts are: HS ode, , , , , and For the last produt group the sae estiation ethod is applied as the one that akes less variation for the rest of produts, but unit prie of this produt group shows ore variation, whose reason is not yet figured out. 10

11 The elastiity estiated for the starter otors et. (HS ) is -1.86, and it is when the ontrating environent variable is inluded in estiation. The agnitude is slightly higher than that of the estiate for the integrated iruits. It shows that there is no big differene between the estiation with and without the ontrating variable, though the estiate of the variable itself is positive and statistially signifiant at the 5 % level. The estiate for the safety seat belts et. is obtained to be and for odels without and with the ontrating environent variable, respetively. There is soe differene between the two estiates, however, the estiate of the variable itself is not statistially signifiant, though positive. Sine the agnitude of the elastiity is relatively high, it ight be inferred that the part provided by Japanese suppliers are not so different fro the part provided by other ountries. These estiation results iply that U.S. iports of soe auto parts, eletroni integrated iruits and starter otors partiularly, signifiantly depend on the ontrating enforeability of the soure ountries. It follows that for those parts the relationship-speifi investent (RSI) is ore required and therefore, the ountries with better ontrating environent and with best prodution tehnique, like Japanese suppliers, are given greater advantage in the U.S., raising the degree of uniqueness of their produt. On the right hand side of Table 2 shows the estiation results for the elastiity of substitution between parts (like safety seat belts and starter otors). The estiates are negative and statistially signifiant. The agnitudes of the estiates are in the sae range (-1.62 ~ -1.95), and relatively low. That is, the parts onsidered are used opleentary in auto prodution as indiated in Proposition (3). It should be notied that the estiate for eletroni integrated iruits shows the lowest agnitude and that only this part estiate shows the intra-lass elastiity of substitution lower than the inter-lass elastiity of substitution. This ight again fortify the ipliation for the industry, naely the RSI is iportant sine both elastiity are low, and one the investent is oitted, it is speifi with other parts as well as to an auto, and hene, the part is less substitutable. 4. Conlusion This study tried to investigate how unique Japanese auto parts and oponents are in a third arket. The siple regression estiation results suggest that the agnitude of the 11

12 estiated elastiities varies aross 18 inputs and aong three destination arkets. Soe parts have higher elastiity of substitution, and others have lower elastiity of substitution. For those inputs with lower elastiity, one ould say that Japanese eports are less likely replaed by eports fro other ountries. But fro this eporter s viewpoint, we do not really know about why soe parts are less substitutable. Then, I estiated the elastiity of substitution with a odel fro outsouring, relationship-speifi investent (RSI), and inoplete ontrat theory. The degree of substitutability depends on the degree of ontratible ativities of the prodution in assoiation with the oitent of RSI on a supplier s side, and all of these ake a partiular part ore unique (speifi). Thus, the ontrat odel is supposed to be better eplaining the pattern of interediate input trade as long as the uniqueness of the input is onerned. Using the two-level CES prodution funtion, I estiated the elastiity of substitution inorporated with the degree of ontrating environent. For the iro ontroller-inluded produt ategories, the estiate of the ontrating environent is positive and signifiant, and the agnitude of the intra-lass elastiity of substitution is lowest. It follows that for this part the RSI is iportant and oitent of the investent by suppliers of a ountry with better ontrat enforeability akes the part speial for akers, but if the part supply were disonneted, the entire auto prodution proess would have to be stopped. This story sees onsistent with the Japanese eperiene after the natural disaster on Marh, The estiation was onduted for the two levels: one is for intra-lass, and the other is for inter-lass. The two elastiities are separately estiated, but whether the differene between the two estiates is statistially signifiant or not was not tested in this study. It is partiularly neessary when arguing on the opleentarity aong inputs in ters of oparison with the intra-lass elastiity. Thus, this test should be onduted as soon as possible. The urrent study also priarily onduted a ross-ountry analysis. Sine tie-series data adds historial inforation, suh data would be probably very iportant espeially when investigating the ontrating nature of interediate input trade, as well as just for the purpose of aking larger saple. Thus, the analysis using a panel data set will also be a net step. 12

13 Referenes Aeoglu, Daron, Pol Antras, and Elhanan Helpan, Contrats and Tehnology Adoption, The Aerian Eonoi Review, Vol. 97, No. 3, June pp Asanua, Banri, Manufaturer-Supplier Relationship in Japan and the Conept of Relation-Speifi Skill, Journal of the Japanese and International Eonois, Vol. 3, No. 1, Marh pp Feenstra, Robert, The International Trade and Investent Progra, Progra Report, NBER Reporter, No. 2, Feenstra, Robert, and Barbara Spener, Contratual Versus Generi Outsouring: The Role of Proiity. NBER Working Paper No.11885, Deeber (August 2006.) Grossan, Gene, and Elhanan Helpan, Outsouring versus FDI in Industry Equilibriu, Journal of the European Eonoi Assoiation, 1, 2003, pp Head, Keith, John Ries, and Barbara Spener, Vertial Networks and U.S. Auto Parts Eports: Is Japan Different? Journal of Eonois and Manageent Strategy, Vol. 13, No. 1, 2004, pp Helpan, Elhanan, International Organization of Prodution and Distribution, NBER Reporter, Suer 2006a. Helpan, Elhanan, Trade, FDI, and the Organization of firs. Journal of Eonoi Literature, Vol. 44, Septeber 2006b, pp Kaufann, Daniel, Aart Kraay, and Massio Mastruzzi, Governane Matters VIII-Aggregate and Individual Governane Indiators , Poliy Researh Working Paper 4978, Developent Researh Group, The World Bank, June Ministry of Eonoy, Trade and Industry, Japan, [METI], White Paper on International Eonoy and Trade Nunn, Nathan, Relationship-Speifiity, Inoplete Contrats, and the Pattern of Trade, Quarterly Journal of Eonois, May pp Nunn, Nathan, and Daniel Trefler, The Boundaries of Multinational Fir: An Epirial Analysis, Helpan, E., D. Marin, and T. Verdier, (eds) The Organization of Firs in a Global Eonoy. Harvard University Press, pp Spener, Barbara, International Outsouring and Inoplete Contrats, Canadian Journal of Eonois, Vol. 38 No. 4, 2005, pp Saito, Mika, Arington Elastiities in Interediate Inputs Trade: A Proble in Using Multilateral Trade Data, The Canadian Journal of Eonois, Vol. 37, No. 4, Noveber 204. pp

14 Sato, Kazuo, A Two-Level Constant-Elastiity-of-Substitution Prodution Funtion, The Review of Eonoi Studies, Vol. 34, No. 2, April pp United Nations, UN Coodity Trade Statistis Database [COMTRADE] (downloaded fro the web site: Yaawaki, Hideki, Japanese Eports and Foreign Diret Investent. Cabridge: Cabridge University Press,

15 Table 1. Elastiity of Eporter Substitution between Japan and Copeting Countries: 18 Auto Parts in U.S., China, and World Markets, 2010 Destination U.S. China World HS 6-digit Code No. siga (σ) t-value R-Squared siga (σ) t-value R-Squared siga (σ) t-value R-Squared ** ** ** ** ** * ** ** * ** * ** ** * * * ** ** ** ** * ** * ** * ** ** Data Soure: United Nations, COMTRADE Database (downloaded fro the web site: Notes: The elastiity of eporter substitution is the estiated oeffiient on ln(wj/w) by OLS with a onstant ter. The asterisks "**" show the 1% signifiane level, and an asterisk "*" shows the 5% signifiane level. The 28 ountries and area are: Australia, Austria, Belgiu, Canada, China, Denark, Finland, Frane, Gerany, Hong Kong, Ireland, Italy, Japan, Luebourg, Malaysia, Meio, Netherlands, New Zealand, Norway, Portugal, Russia, Singapore, South Afria, Spain, Switzerland, Thailand, the U.K., and the U.S. The nuber of observations varies fro 13 to 26 for the U.S. and China, and 14 to 27 for the World. 15

16 Table 2. Elastiity of Substitution for Iported Auto Inputs: 5 HS 6-digit U.S. Iports, 2010 Regression Model Intra-lass elastiity of substitution Inter-lass elastiity of subsitution HS 6-digit elas. of sub. (σ ) ontrating environent Adjusted elas. of sub. (σ) Code No. ln(w /W J ) t-value ln(r /R J ) t-value R-Squared No. of Obs. ln(p n /P ) t-value R-Squared No. of Obs ** * ** ** ** ** ** ** ** ** ** * ** Data Soure: United Nations, COMTRADE Database (downloaded fro the web site: Notes: Due to the data estiation proble by the UN, we do not have an enough variation of unit pries aong ountries based on the alulation by the value-divided-by-quantity ethod. Those produts with a virtually fied unit prie are eluded in this analysis, liiting to the five parts out of 18 in Table 2. Estiation inludes a onstant ter. The asterisks "**" show the 1% signifiane level, and an asterisk "*" shows the 5% signifiane level. 16

17 Appendi Table to Tables 1 and 2: HS 6-digit Classifiation Codes and Produt Naes (UN, COMTRADE Database) : "Parts suit. for use solely/prinipally with the ahines of 85.01/85.02","Parts suitable for use solely or prinipally with the ahines of heading or ",1,6,"8503" : "Starter otors & dual purp. starter-generators","- Starter otors and dual purpose starter-generators",1,6,"8511" : "Parts of the equip. of 85.12","- Parts",1,6,"8512" : Eletroni integrated iruits, proessors & ontrollers, whether/not obined with eories, onverters, logi iruits, aplifiers, lok & tiing iruits,/other iruits : "Chassis fitted with engines, for the otor vehiles of ","Chassis fitted with engines, for the otor vehiles of headings to ",1,6,"8706" : "Bupers & parts thereof of the otor vehiles of ","- Bupers and parts thereof",1,6, "8708" : "Safety seat belts of the otor vehiles of ","-- Safety seat belts",1,6, "8708" : "Parts & aessories of bodies (inl. abs) of the otor vehiles of , n.e.s. in 87.08","-- Other",1,6,"8708" : "Brakes & servo-brakes; parts thereof, of the otor vehiles of headings to ","- Brakes and servo-brakes; parts thereof",1,6,"8708" : "Gear boes & parts thereof, of the otor vehiles of headings to ","- Gear boes and parts thereof",1,6,"8708" : "Drive-ales with differential, whether/not provided with other transission oponents, & non-driving ales; parts thereof of the otor vehiles of headings to ", - Drive-ales with differential, whether or not provided with other transission oponents, and non-driving ales; parts thereof,1,6,"8708" : "Road wheels & parts & aessories thereof for the otor vehiles of ","- Road wheels and parts and aessories thereof",1,6,"8708" : "Suspension systes & parts thereof (inl. shok-absorbers),of the otor vehiles of headings to 87.05", - Suspension systes and parts thereof (inluding shok-absorbers),1,6,"8708" : "Radiators & parts thereof for the otor vehiles of ","-- Radiators and parts thereof",1,6,"8708" : "Sileners (ufflers) & ehaust pipes; parts thereof for the otor vehiles of ","-- Sileners (ufflers) and ehaust pipes; parts thereof",1,6,"8708" : "Cluthes & parts thereof for the otor vehiles of ","-- Cluthes and parts thereof",1,6,"8708" : "Steering wheels, steering oluns & steering boes; parts thereof for the otor vehiles of ", -- Steering wheels, steering oluns and steering boes; parts thereof,1,6,"8708" : "Other parts & aessories for the otor vehiles of , elud /92/93/94/95.","-- Other",1,6,"8708" 17