COMPILATION AND APPLICATIONS OF IDE-JETRO'S INTERNATIONAL INPUT OUTPUT TABLES

Transcription

1 Economc Systems Research ISSN: (Prnt) (Onlne) Journal homepage: COMPILATION AND APPLICATIONS OF IDE-JETRO'S INTERNATIONAL INPUT OUTPUT TABLES Bo Meng, Yaxong Zhang & Satosh Inomata To cte ths artcle: Bo Meng, Yaxong Zhang & Satosh Inomata (2013) COMPILATION AND APPLICATIONS OF IDE-JETRO'S INTERNATIONAL INPUT OUTPUT TABLES, Economc Systems Research, 25:1, , DOI: / To lnk to ths artcle: Publshed onlne: 21 Mar Submt your artcle to ths journal Artcle vews: 251 Vew related artcles Ctng artcles: 12 Vew ctng artcles Full Terms & Condtons of access and use can be found at Download by: [Asa Keza Kenkyusho] Date: 15 September 2015, At: 19:22

2 Economc Systems Research, 2013 Vol. 25, No. 1, , COMPILATION AND APPLICATIONS OF IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES BO MENG a *, YAXIONG ZHANG b and SATOSHI INOMATA c a Insttute of Developng Economes (IDE) JETRO, Chba, Japan; b Department of Economc Forecastng, State Informaton Center, Bejng, Chna; c Insttute of Developng Economes (IDE) JETRO, Chba, Japan (Receved 24 July 2012; In fnal form 19 December 2012) Internatonal nput output (IO) tables are among the most useful tools for economc analyss. Snce these tables provde detaled nformaton about nternatonal producton networks, they have recently attracted consderable attenton n research on spatal economcs, global value chans, and ssues relatng to trade n value added. The Insttute of Developng Economes at the Japan External Trade Organzaton (IDE-JETRO) has more than 40 years of experence n the constructon and analyss of nternatonal IO tables. Ths paper explans the development of IDE-JETRO s mult-regonal IO projects ncludng the constructon of the Asan Internatonal Input Output table and the Transnatonal Interregonal Input Output table between Chna and Japan. To help users understand the features of the tables, ths paper also gves examples of ther applcaton. Keywords: nternatonal nput output; mult-regonal; producton networks 1. INTRODUCTION The Insttute of Developng Economes at the Japan External Trade Organzaton (IDE- JETRO) has constructed the Asan Internatonal Input Output (AIIO) table for the years 1975, 1985, 1995, 2000, and These datasets are recognzed as a powerful analytcal tool for the study of ndustral structure and producton networks across countres and are used worldwde. In the last few years, however, some promnent academc nsttutons and nternatonal organzatons have launched projects amed at constructng mult-regonal nput output (MRIO) tables, as shown n other papers n ths volume. Ths reflects not only an ncreased demand for nternatonal nput output (IIO) data, but also mples a need for dfferentaton n the products n order to meet a varety of user needs. One way for an MRIO project to dfferentate tself s to broaden ts coverage of countres n order to appeal to a wder range of users. However, ths entals sgnfcant costs, because the more countres the table covers, the more dffcult t wll be to harmonze consttuent natonal tables nto a common format. Another strategy s to focus on provdng more specfc regonal nformaton for a lmted number of countres. Insttute of Developng Economes (IDE) pursues ths drecton wth *Correspondng author. E-mal: mengbo@de.go.jp, bo_meng@de.go.jp 1 A prelmnary verson of the 2005 AIIO table was completed n It wll be offcally released n The Internatonal Input Output Assocaton

3 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 123 the new Transnatonal Interregonal Input Output (TIIO) table. Ths project seeks to lnk the nterregonal IO tables of several countres nto a sngle matrx usng regonal mport/export data for ndvdual countres. Ths paper explans the development of IDE s MRIO projects and ther evoluton along these two dfferent yet complementary approaches. Sectons 2 and 3 descrbe the AIIO and TIIO projects and the basc features of the tables. In Secton 4, we explore two applcatons of these datasets: usng the AIIO table to measure trade n value added (TVA) n the Asa- Pacfc regon and usng the TIIO table to estmate the cross-natonal transfer of employment opportuntes. Secton 5 dscusses the challenges faced when seekng to upgrade and enhance the database, and Secton 6 closes the paper by referrng to the recent drectons of IDE s MRIO projects. 2. COMPILATION OF AIIO TABLES 2.1. Hstory of IDE s AIIO Project The hstory of IDE s AIIO project can be dvded nto four phases. Durng the frst phase (before 1973) several poneerng works explored the basc concepts of IIO tables. In 1964, Watanabe (1966) proposed the dea of usng IIO models to analyze the North-South trade problem. Buldng upon ths dea, n 1965, IDE developed an IIO model coverng sx regons: North Amerca, Europe, Oceana, Latn Amerca, Asa, and Japan. In 1966 and 1971, IDE constructed IIO models for 10 Asan countres. These tables were subject to a number of lmtatons, however, such as smple estmaton technques and rough ndustry sector classfcaton. In the second phase ( ), IDE developed a comprehensve IIO table to explore nter-ndustry transactons among East Asan countres. In 1973, the Republc of Korea and the fve ASEAN countres plus the USA were chosen to be the endogenous member countres, because these countres have close economc relatonshps wth Japan. Had all the countres compled ther natonal IO tables usng a common reference year, the project would have been much easer. At the tme, however, Indonesa, Thaland, and Sngapore had not yet constructed offcal IO tables. IDE also lacked the experence necessary to comple a comprehensve IO table. Thus, two prelmnary tasks were necessary before the project could begn. The frst was to construct natonal IO tables for the three countres that lacked them, and the second was to comple blateral IO tables wth the countres for whch the natonal tables were already avalable. Under ths project, three natonal IO tables (Indonesa for 1971, Sngapore for 1973, and Thaland for 1975) and three blateral IO tables (Korea Japan for 1970, USA Japan for 1970, and Phlppnes Japan for 1970) were constructed n collaboraton wth the natonal statstcal offces and research nsttutes of the countres concerned. In 1978, IDE entered the thrd phase ( ) of the IO project, wth the am of constructng a multlateral IO table among the ASEAN countres, Japan, Korea, and the USA for the year Ths project conssted of three prncpal steps. Frst, exstng tables were updated to the year 1975 for Malaysa, the Phlppnes, Sngapore, and the USA. Second, the 1975 blateral IO tables for Indonesa Japan, Thaland Japan, and Korea Japan were constructed. Fnally, these natonal and blateral tables were lnked together nto a sngle multlateral IO table for Ths table was completed n The 1975

4 124 B. MENG et al. multlateral table has been used for varous analyses of East Asan and Southeast Asan ndustral structure. It also became the prototype for the subsequent IIO projects. After the completon of the 1975 IIO tables, IDE soon began the constructon of the 1985 IIO table wth the am to expand coverage to more Asa-Pacfc countres. Chna had rapdly ncreased ts external trade wth the USA, Japan, and others snce the commencement of the Open Door Polcy, and had come to play an mportant role n the Asa-Pacfc regon, not only as a ggantc market but also as a recpent of nvestments from neghborng countres. Thus, Chna and Tawan were covered n the 1985 multlateral table, makng t even more comprehensve than the 1975 verson. Snce then, IDE has compled multlateral tables every 5 years, provdng powerful tools for analyzng dynamc structural changes n the Asa-Pacfc regon. IDE s IIO projects nvolve many partcpatng organzatons from dfferent economes. Most of these organzatons are governmental or quas-governmental statstcal nsttutes. More than 70 experts from 10 dfferent economes have contrbuted to the complaton of the 2000 AIIO table Complaton Procedure and Methodology of the AIIO Table The AIIO table wth 10 endogenous economes s outlned n Appendx 1. The 76 ndustry classfcaton s gven n Table A1 n a separate supplementary document that s accessble through the webste of ths journal. Broadly speakng, the complaton process of the AIIO table goes through three dstnctve phases. What follows s a step-by-step llustraton of how the AIIO table s compled Adjustment of Presentaton Format Despte the fact that IO tables consttute the central apparatus of the System of Natonal Accounts (SNA), the natonal tables for the ndvdual countres have dfferent features and characterstcs, reflectng each country s economc dosyncrases and avalablty of data. Ths presents dffculty when complng IIO tables from each natonal IO table, snce the data should be mutually consstent and comparable between any parts of the whole. Accordngly, one of the most complcated tasks of complaton s the adjustment of the natonal tables to a common format. In general, detaled, nformaton-rch tables must be conformed to the lessdetaled ones, as the other way round would requre costly efforts to obtan addtonal data. Trade-offs between the level of unformty and the level of nformaton must, therefore, be made, and hence careful and thorough consderaton s necessary when makng adjustments. In , a survey was conducted n order to construct a basc set of nformaton to ad n the desgn of the AIIO common format and adjustment rules. Ths survey was desgned to capture every mportant aspect of an IO table. The questons are grouped nto seven broad categores: benchmark year and recordng prncples, avalablty of natonal 2 The collaboratng nsttutons n the 2000 AIIO project are: the State Informaton Center n Chna; Statstcs Indonesa; the Bank of Korea; the Department of Statstcs n Malaysa; the Natonal Statstcs Offce n the Phlppnes; Busness Research Consultants n Sngapore; the Tawan Research Insttute; the Natonal Economc and Socal Development Board n Thaland; the Appled Research Insttute, Inc. n Japan, and IDE-JETRO. In addton, some statstcal experts from Gfu Shotoku Gakuen Unversty (Prof. Sano Takao) and the Mnstry of Economy, Trade, and Industry of Japan also provded ther help n ths project.

5 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 125 tables and supportng tables, valuaton, form and coverage, specal treatment, publc or sempublc sectors, and response to the 1993 SNA. Based on the survey results, the dfferent features of natonal IO tables across countres can be summarzed as shown n Table 1 (for detaled explanaton, see Appendx 2) Preparaton of Sector Concordance and Supplementary Data (1) Preparaton of sector concordance Each natonal table has ts own ndustry classfcaton system. In the case of the benchmark tables for the 2000 AIIO table, the number of ndustry sectors ranges from 98 for the Malaysan table to 517 for the Japanese table. The weghts of the ndustry categores also dffer. Countres wth large agrculture-based economes have relatvely detaled classfcaton of agrcultural sectors, whereas ndustralzed economes have more comprehensve coverage of manufacturng sectors. As such, the sector classfcaton reflects the characterstcs of the economy concerned, and a precse converson system that brdges natonal codes and AIIO codes s absolutely essental for the complaton of consstent IIO tables. The system of sector concordance has a tree-lke structure, where AIIO classfcaton (the broadest category) s located at the top, and each AIIO code corresponds to one or several natonal codes. The natonal codes are subcategorzed nto the Harmonzed System (HS) of Foregn Trade Statstcs, whch may be further converted to the Standard Internatonal Trade Classfcaton (SITC), another classfcaton system for trade data. If the concordance system has such a clear tree structure, the aggregaton of natonal tables nto AIO classfcaton poses no dffculty. The problem arses when a natonal code s assocated wth more than one AIIO codes. For example, Sngapore s natonal code SIO092 Land transport equpment corresponds to both AIO055 Motor vehcles and AIO056 Motorcycles. We must, therefore, splt the sector at the natonal IO table level before the aggregaton procedure. (2) Supplementary data For the complaton of nternatonal tables, each country must prepare the followng supplementary data accordng to the AIIO sector classfcaton: (a) mport data by commodty and economy of orgn; (b) export data by commodty and economy of destnaton; (c) mport dutes and commodty taxes by commodty; (d) domestc trade margns and freght transport costs (transportaton and trade margn (TTM)) on exported goods by commodty; (e) nternatonal freght and nsurance data by commodty and economy of orgn; and (f) other relevant nformaton. The mport and export data can be drectly constructed from the Foregn Trade Statstcs wth the help of the HS (or SITC) natonal IO AIIO sector concordance. The data for mport dutes and mport commodty taxes, on the other hand, are ndependently presented n the orgnal natonal IO tables n most cases, but f not (as n the case of the US table), they must be collected from the Foregn Trade Statstcs. The data on TTM for exports comes from the supportng tables of the natonal IO tables. Ideally, costs for exported goods (for the delvery from factores to ports) should be used, but f they are not avalable from the table, the average fgures of the TTM matrces can be used as proxes.

6 TABLE 1. Dfferent features n natonal IO tables of the 2000 AIIO member economes. Chna Indonesa Japan Korea Malaysa Tawan Phlppnes Sngapore Thaland USA 1. Converson of valuaton 1.1 Basc prce to producer s prce 1.2 Prvate consumpton expendture 1.3 Export vectors 1.4 Import matrx/vector 2. Negatve entres 3. Dummy sectors 4. Machne repar 5. Fnancal ntermedares 6. Specal treatment of mport/export 6.1 Water transportaton 6.2 Pure mport of gold 6.3 Re-export 6.4 Telecommuncaton 7. Computer software products 8. Producers of government servces 126 B. MENG et al.

7 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 127 Fnally, data on nternatonal freght and nsurance costs are collected from the Foregn Trade Statstcs, where avalable. However, snce these data are not avalable for all countres, we use a gravty model to estmate the mssng nformaton. In our model, two measures of dstance are calculated: the shppng-route dstance and the straght-lne dstance. The shppng-route dstance s taken from the Dstance Tables for World Shppng (JSE, 1983) whch reports the dstances between major ports. The straght-lne dstance, whch can be regarded as an analog of the ar-flght dstance, s calculated between commercal centers of the countres concerned. Of these two measures, the one that better explans varaton n nternatonal freght and nsurance rates s used. 3 It should be noted that the qualty of the mport matrces plays a crtcal role n determnng the accuracy of the IIO table. To ncrease the accuracy of the mport matrces, a specal survey (see Inomata and Kuwamor, 2008) on mported commodtes has been conducted n the current AIIO project. The man goals of the survey are as follows: (a) to dentfy whch ndustres utlzed mported commodtes by orgn country; (b) to determne the cost of nternatonal freght and nsurance on each mported commodty; and (c) to determne the cost of mport dutes and commodty taxes leved on each mport commodty. The survey respondents are establshments that mport the commodtes (manufacturers, tradng frms, etc.), snce they are expected to possess nformaton on amounts mported by country of orgn and the amount dstrbuted to domestc ndustres. The survey s carred out as an ndependent sample survey. It may also be conducted as a rder survey attached to other offcal surveys, whch s more effcent and comprehensve. A sample form of the questonnare descrbed above s presented n Table A2 n the supplementary nformaton on the webste of ths journal Lnkng and Balancng An IIO table s not just a patchwork of the peces taken from natonal tables, but rather a product of careful complaton of supplementary data and manual reconclaton. Ths secton gves a bref descrpton of the lnkng and balancng work of the AIIO tables. (1) Lnkng of natonal IO tables All the peces of each member country prepared n the prevous steps are lnked together to form the table as shown n Appendx 1. The basc dea of lnkng s to replace the export vector wth the mport matrx of the tradng partner. In other words, we frst splt the mport matrx of each country s natonal IO table nto all ts partner countres and then lnk these tables row-wse. At ths stage, the valuaton of mports n each country s natonal IO table s also converted from the CIF (cost, nsurance, and freght) prce to the producer s prce by usng data on nternatonal freght and nsurance and domestc transport costs and trade margns compled n the prevous steps. (2) Balancng and reconclaton of the fnal table 3 For detaled nformaton on the estmaton results of gravty models, refer to Kuwamor (2006).

8 128 B. MENG et al. FIGURE 1. Balancng and reconclaton procedures. Note: CT, the fgure of output by sector s used as the control total n our adjustment procedure. Start - (a) Perform macrocheck wth offcal data (b) Compute control total (CT) s error rate (c) CT s error rate > specfed crteron? Yes No End (d) Calculate the error n trade statstcs (e) Identfy the reasons for errors (f) Make adjustment polcy/nstructon (g) Run adjustment program The fnal step of complaton s the manual balancng and reconclaton work. The table s balanced wth respect to nput composton, but total demand s not necessarly consstent wth total supply for each country at ths stage. Such an mbalance stems from several factors. The frst s the nconsstency n sector classfcatons between countres. Though each country s requred to make ts own code concordance from the HS code to the AIIO sector classfcaton, the possblty of dfferences n statstcal concept stll exsts. The second factor s entrepôt trade. Ths type of trade s often counted n dfferent ways by trade partners. For example, exports va Hong Kong to the USA may be counted by the USA as mports from Chna. In the case of Sngapore, where nternatonal trade s extremely large compared wth the scale of the economy, and there s a large volume of entrepôt trade, there are especally large statstcal dscrepances n the related nternatonal trade matrces. The balancng procedure shown n Fgure 1 s repeated untl the results satsfy the specfed crtera. Addtonal spot checks are conducted at the end of the adjustment to fnd any unusual entres n the table that mght have been made durng the adjustment process. 3. COMPILATION OF THE TIIO TABLE BETWEEN CHINA AND JAPAN After the enactment of the Open Door Polcy n 1978 and Chna s accesson to the WTO n 2001, the ncreasng economc presence of Chna has brought sgnfcant changes n the ndustral network between Chna and Japan. Wth the contnuous reducton of nternatonal trade costs, the degree of economc nterdependence of Chna and Japan has grown tremendously n recent years. At the same tme, the cross-border economc nteracton between the two countres also affects each country s domestc economes at the regonal level. Snce

9 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 129 both Chna and Japan have relatvely large economc scales, the economy of just one of ther domestc provnces may be larger than that of a md-sze country. We should, therefore, consder the possblty of provdng nformaton measurng the transnatonal nterregonal economc mpacts of trade between Chna and Japan. To explore ths, the IDE-JETRO launched a new nternatonal jont project (see IDE, 2006) n 2006 to comple the TIIO table for Chna and Japan, lnkng the ndvdual nterregonal IO tables of the two countres Overvew of the TIIO Table The 2000 TIIO table for Chna and Japan depcts the transnatonal, nterregonal, and nterndustry networks extended over 18 regons. 4 The regons covered nclude seven domestc regons of Chna, eght domestc regons of Japan, ASEAN5 (.e. Indonesa, Malaysa, the Phlppnes, Sngapore, and Thaland aggregated nto one regon), East Asa (South Korea and Tawan aggregated nto one regon), and the USA. The domestc regons of Chna and Japan are aggregatons of ther domestc provnces. Appendx 3 shows the detaled nformaton for the regons. Due to tme and budget lmtatons, the TIIO project was desgned to be completed wthn 1 year. Lmted data avalablty and a desre for smplcty necesstated the use of relatvely rough ndustry classfcatons (10 sectors) for ths project. Instead of usng a tradtonal classfcaton, such as lght ndustry versus heavy ndustry, the followng 10 sectors are employed n the TIIO table, consderng recent developments n producton technologes: (1) Agrculture, lvestock, forestry, and fshery; (2) Mnng and quarryng; (3) Household consumpton products; (4) Basc ndustral materals; (5) Processng and assemblng; (6) Electrcty, gas, and water supply; (7) Constructon; (8) Trade; (9) Transportaton; and (10) Servces. The 2000 TIIO table s shown n Appendx 4. The layout s smlar to that of the AIIO table shown n Appendx 1. Ths table provdes detaled transnatonal, nterregonal, and nter-ndustry nformaton on the nput composton and output dstrbuton of each domestc ndustry vs-à-vs home and foregn countres ndustres Complaton Procedure and Methodology of the TIIO Table The complaton process of the TIIO table s very smlar to that of theaiio tables. The complaton conssts of three prncpal steps. The frst step s to reconcle the dfferences n format between the Chnese and Japanese sdes. The second step s to prepare the transnatonal and nterregonal trade data and splt the mport matrces of Chna and Japan s domestc nterregonal IO tables by regon of orgn. The thrd step s to lnk the two tables completed n the second step and to reconcle any dscrepances Data Collecton and Estmaton The man sources of data used n the TIIO project nclude (1) the 1997 nterregonal IO table of Chna; (2) the 2000 nterregonal IO table of Japan; (3) the 2000 AIIO table; 4 The 2000 TIIO project s manly an nternatonal collaboraton between the State Informaton Center (SIC) n Chna and the Appled Research Insttute, Inc. (ARI) and IDE-JETRO (project manager) n Japan. In addton, some experts from the Natonal Insttute for Land and Infrastructure Management (NILIM), the Mnstry of Land, Infrastructure, and Transport of Japan, and Pacfc Consultants Co,. Ltd are nvolved n ths project as consultants and advsors.

10 130 B. MENG et al. (4) the 2000 commodty flow data by port between Chna and Japan; and (5) other supplementary data. The 2000 AIIO table s used as control data for the project. The SIC and ARI are n charge of the complaton of the Chnese and Japanese components, respectvely. Both parts are nterregonal IO tables wth a separate mport matrx for 18 endogenous regons. To construct these tables, both nsttutons use nformaton on commodty flow by port between Chna and Japan to splt the mport matrx of ther nterregonal IO tables. The commodty flow data by port are estmated by consultants at NILIM Complaton Methodology Snce there are nsuffcent survey data for us to make detaled estmates of transnatonal and nterregonal commodty flows, some relatvely strct assumptons are used. The man complaton method s based on the followng prncple. Suppose an nternatonal (blateral) IO table between country R and country S and an nterregonal IO table between regon p and regon q of country R. Country R n the IIO table s dvded nto two regons (p and q) by usng nformaton obtaned from the nterregonal IO table and transnatonal commodty flow data by port. To dvde the row-wse transactons of country R nto regons, the regonal rato of the nterregonal IO table s appled to the domestc transacton and the regonal export rato s used to dvde the export matrx for country S. Thus, the output structure of regon p n country R can be estmated as follows: (z pp j (f pp l (e ps + z pq j + f pq l e ps + e qs e ps z pp j + z qp j f pp l + f qp l + z qq j ) zrr j = ntermedate output of regon p n country R, + f qq l ) f l RR = fnal demand of regon p n country R, ) zrs j = exports from regon p n country R to ntermedate sectors n country S, (e ps + e qs ) f l RS = exports from regon p n country R to fnal demand n country S, x p (x p + x q )xrr = gross output of regon p n country R. Here, z, f, e, and x represent ntermedate, fnal products, exports, and output, respectvely; (j) and l represent ndustry and fnal demand tems. Regon q n country R s also estmated n the same manner. After splttng the row vector of country R nto regons, the columnwse dvson s done n the same way as descrbed above. The nput structure of regon p n country R s estmated as follows: (m Sp (m Wp m Sp + m Sq m Wp + m Wq ) zsr j = mports of regon p n country R from country S, ) zwr j = mports of regon p n country R from the rest of the world (W),

11 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 131 v p kj (v p kj + vq kj )vr kj = value added of regon p n country R, x p j (x p j + x q j )xr j = gross nput of regon p n country R. Here, m represents mports; v represents value-added tems. Fnally, mport matrx for trade to regon p from country S s dvded by regon of orgn n country S. Lettng the regon of country S be y, we then have e yr y eyr (m Sp m Sp + m Sq ) zsr j = mports of regon p n country R from regon y n country S. Imbalances can arse n both rows and columns. Adjustments are made based on the RAS method and experts opnons n order to elmnate the dscrepances. 4. APPLICATION EXAMPLES OF IDE S AIIO AND TIIO TABLES 4.1. TVA n the Asa-Pacfc Regon The ncreasng sgnfcance of global value chans (GVCs) s among the most mportant features of rapd economc globalzaton. Phenomena related to GVCs nclude vertcal specalzaton, fragmentaton of producton, outsourcng, and global supply chans. Despte the dfferent terms used, all of them ndcate that a hgher volume of ntermedate goods, such as parts and components, s produced n sequental stages or processes across dfferent countres and then exported to other countres for further producton. Ths phenomenon s explaned by the so-called second unbundlng (Baldwn, 2011). Baldwn explans how lower communcaton costs resultng from the nformaton and communcaton technology revoluton have enabled the nternatonal unbundlng of factores and offces, whch means that tasks as well as products can be traded globally. In an attempt to understand the evoluton of GVCs among countres, many ndcators based on trade statstcs have been developed. However, most of these statstcs focus on showng the drect trade-specfc relatonshps between countres, rather than consderng the roles that nter-country and nter-ndustry producton networks play n the global economy. Ths secton uses the concepts of TVA 5 based on the AIIO tables to show an example of measurng GVCs n the Asa-Pacfc regons. The TVA metrc shows how much of a country s (country A s) value added s nduced by another country s (country B s) fnal demand both drectly and ndrectly through nternatonal producton networks. Ths can also be regarded as a country s (country A s) export of value added to the other country (country B). Fgure 2 shows the TVA of the Asa- Pacfc regon by country of orgn and destnaton for 1985 and That s, each bar gves the mports of value added by one of the countres and the bar s splt nto the nne 5 For detaled nformaton of the concept of TVA, refer to the jont report by WTO-IDE (2011) and OECD-WTO (2012).

12 132 B. MENG et al. FIGURE 2. Trade n value added n the Asa-Pacfc regon (1985, 2005) Mllon US$ n nomnalterms USA Thaland Sngapore Phlppnes Tawan Malaysa Korea Japan Indonesa Chna 10, , , , , , , ,000 Chna Indonesa Japan Korea Malaysa Tawan Phlppnes Sngapore Thaland USA countres of orgn from where the value added s exported. The man features of the fgure are summarzed below. (1) In 1985, mports of value added by Japan and the USA domnate the Asa-Pacfc regon s GVCs at the absolute level. However, n 2005, Chna s mports of value added are the largest followed by those of the USA and Japan. Ths reflects the rapdly ncreasng potental of Chna to be a growth engne of global trade through ts value-added gan potental. (2) South Korea and Tawan s presence n the gan of value added from regonal markets also ncreased rapdly. A smlar phenomenon can also be found for the other Asan economes, such as Thaland, Malaysa, and Sngapore. Ths reflects the ncreasng partcpaton of Asan economes n the value chans of the Asa-Pacfc regon between 1985 and (3) When lookng at the components of the TVA fgures by country, we can see that, n 1985, the fnal demands of the USA and Japan were the man sources n the value-added creaton processes of almost allasan countres. In 2005, however, Chna exported much more value added, especally to the East Asan economes (Japan, South Korea, and Tawan). Ths ndcates that Chna s not smply a world factory or supply center, but also a demand center at least for the Asan regon. Applyng the TVA metrc to the AIIO tables makes t possble to capture the creaton and dstrbuton of value added n the Asa-Pacfc regon. Ths knd of applcaton can avod the double countng that occurs when trade data are used to measure GVC-related phenomena. In addton, the TVA concept may change the tradtonal measure of the blateral trade balance. Ths can help us understand the country-to-country trade relatonshp wthn the

13 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 133 GVCs n detal. For a more detaled example of ths applcaton, refer to the WTO-IDE (2011) jont report Transnatonal and Interregonal Impacts of Japan s Fnal Demand on Chna s Employment 6 Chna and Japan are both mportant world economes and there are very strong trade lnkages between them. Before 2003, Japan had long been Chna s largest trade partner, and n the 1990s, Chna was Japan s second largest trade partner, just behnd the USA. The rapd development of Chna s economy and the expanson of ts trade lnkages wth Japan can beneft employment n both countres sgnfcantly. Chna s a country wth a large employment base and a great trade power, so there s mportant practcal sgnfcance n studyng the effects of foregn trade on ts employment. However, most exstng research treats Chna as a whole rather than focusng on ts domestc regons. Because of the great varaton n economc sze, ndustral structure, producton effcency, and overseas dependency across regons wthn Chna, there s a need for more regonal level perspectves to help to mprove understandng of the creaton and dstrbuton of job opportuntes caused by foregn trade. In addton, regonal level analyses provde mportant nformaton and reference ponts for local governments, who are the actual executors of the central government s employment creaton polces. The TIIO model between Chna and Japan for the year 2000 provdes tools to analyze the effects of trade between Japan and Chna on employment for dfferent domestc regons and ndustres. Based on the model, we can see that n 2000, Chna s domestc fnal demand contrbuted mllon jobs, accountng for 90.8% of the total employment n Chna. Ths also means that external demand created 9.2% or 65.3 mllon jobs for Chna, and Japan s contrbuton accounts for 2.25% or mllon jobs n Chna. Fgure 3 shows, by ndustry and regon, the employment n Chna nduced by Japan s fnal demand n The man features are summarzed below. (1) The three coastal regons (East Coast, North Coast, and South Coast) are the largest benefcares followed by the Northeast and Central regons. The coastal regons are the most developed areas wth large economes and strong trade lnkages wth Japan, so ther poston at the top s not surprsng. The East Coast and South Coast regons are located downstream n Chna s domestc supply chans (see Meng et al., 2012); therefore, a worker n the household consumpton product ndustry may fnd greater job opportuntes because of trade wth Japan. In addton, a number of large free trade zones and export processng zones are located n these two coastal regons. We, therefore, see that employment n the processng and assemblng sector accounts for a relatvely large share of total employment compared to other regons. (2) In the North Coast, North East, and Central regons, the agrculture sector receves the largest beneft from Japanese demand. These regons not only have a relatvely large agrcultural ndustry wth the ablty to export drectly to Japan, but also occupy upstream postons n the domestc supply chans. The coastal regons exports of 6 Thanks to Ms An Peng, Central Unversty of Fnance and Economcs, Chna, for her dlgent assstance on ths secton (for detals, see Zhang et al., 2012).

14 134 B. MENG et al. FIGURE 3. Regonal employment n Chna nduced by Japan s fnal demand. 5,000 Thousand persons Urban hgh-sklled employment 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000 By educaton level By ndustry Urban low-sklled employment Uural non-agrcultural employment Agrcultural employment Servces Transportaton Trade Constructon Electrcty, gas, and water supply Processng and assemblng Basc ndustral materals Household consumpton products Mnng and quarryng Agrculture Northeast North coast East coast South coast Central Northwest Southwest household consumpton products (especally food products) depend on the supply of agrcultural products from the agrculture-orented regons such as the North Coast, North East, and Central regons. Ths knd of backward lnkage through both global supply chans (Japan s demand for Chna s food products produced n the coastal regons) and Chna s domestc supply chans (the coastal regons demand for agrcultural products provded by other nner regons whch are then used to produce food products for Japan) can be consdered a prmary reason that more than half of the employment creaton n the North Coast, North East, and Central s from the agrculture ndustry. (3) When lookng at the nduced regonal employment by the educaton level, we note that urban low-sklled workers and rural non-agrcultural workers account for large shares of the total employment created by Japan s fnal demand. Ths ndcates that Chna s exports to Japan are labor-ntensve and requre lttle educaton to produce. The above result s a smple applcaton of the TIIO table. For more detaled applcaton examples at the transnatonal and nterregonal level, one can refer to Shbasak et al. (2008) and Inomata and Meng (2011). 5. FUTURE CHALLENGES IN THE PRODUCTION OF IIO TABLES 5.1. Tme Lag of Data Release IDE comples the AIIO tables every 5 years. However, there s always a tme lag of more than 5 years between the benchmark year and the reference year. Snce most countres construct ther natonal IO table every 5 years, and the benchmark years across countres dffer, ths makes t dffcult to speed up the process of lnkng the countres tables. If more Asan countres utlzed the Supply and Use Table (SUT) and the annual natonal IO table, the complaton tme of the IIO table could be greatly reduced.

15 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES Dscrepances Caused by Conflctng Data As mentoned n the prevous sectons, the man reasons for dscrepances n the data are (1) the nconsstency of export/mport fgures between the natonal IO table and nternatonal trade statstcs; (2) the mrror statstcs problem n blateral trade statstcs caused by how re-export and re-mport are treated; (3) dfferences n the valuaton of export statstcs (at FOB prces) and mport statstcs (at CIF prces). One possble soluton to the above problem (1) s to carefully consder the consstency among IO-based trade data, BOP statstcs, and custom-based trade statstcs. In addton, the Unted Natons Statstcs Dvson (UNSD) has proposed new recommendatons for the Internatonal Merchandse Trade Statstcs, accordng to whch mports wll be reported on the FOB bass n addton to the standard CIF valuaton. Ths should help consstency wth regard to blateral trade data. Fnally, the re-export statstcs by country of orgn and destnaton should help to allevate the mrror statstcs problem Valuaton System The AIIO tables are shown wth producer s prces snce most of our member countres construct ther natonal IO tables usng producer s prce. In recent years, many Asan economes have consdered establshng or mprovng ther SUT system under an nternatonal jont project supported by the Asan Development Bank. If SUT data for more Asan countres becomes avalable, the natonal IO table at basc prces can be estmated more easly. 6. RECENT DIRECTIONS OF OUR MRIO PROJECTS 6.1. Complaton of Chna s Mult-regonal IO Tables SIC and IDE jontly compled the wdely used 1997 Chna 8-regon, 30-ndustry multregonal IO (CMRIO) model (see IDE, 2003). Snce then, Chna s nterregonal trade and economc lnkages have strengthened, creatng an urgent need to update and develop the 2002 and 2007 CMRIO model for the complaton year of Chna s IO table. Drawng upon the experence of developng the 1997 CMRIO model, SIC has developed and mproved methods and procedures. Ths resulted n 2002 and 2007 MRIO models that were based on more normatve methods. In the constructon of the 2002 and 2007 CMRIO models, SIC adopted the ndustry classfcaton system used by the Natonal Bureau of Statstcs (NBS). Some of the servce ndustres were aggregated so that the models are now classfed at 29 ndustry levels. Ths ndustry classfcaton methodology s bascally comparable wth that used for the 1997 CMRIO model, allowng for the establshment of a seres of CMRIO models. The regonal dvsons used for the 2002 and 2007 models are dentcal to the ones used for the 1997 model. All research and development work for the 2002 and 2007 CMRIO models, however, s based on separate nformaton on 30 provnce-level dvsons (provnces, muncpaltes, and autonomous regons), 7 allowng for dfferent combnatons n regonal dvson and thus the establshment of CMRIO models for dfferent regons. 7 Snce the IO table for the provnce of Tbet s not compled, t s not ncluded n provnces descrbed n ths paper. The economc structure of Tbet s also not consdered n the development of the CMRIO model.

16 136 B. MENG et al. There are several unque characterstcs of the 2002 and 2007 CMRIO model development. Frst, snce the entre development s based on 30 provnces, varous regonal versons of the CMRIO model can be made after aggregatng dfferent provnce groups accordng to the needs of the regonal plan and the polcy desgn. Second, SIC proposes specfc models and methods based on entropy maxmzng and gravty models for estmatng the nterregonal trade coeffcent. Thrd, n cooperaton wth NBS, SIC modfes the nterregonal transacton matrx by usng the basc survey data on nter-provnce nflow and outflow as found n Natonal Input Output Survey Fourth, when conductng balance adjustment, each provnce s table s fully utlzed to ensure that the sum of all provnces parts equals the totals found n the natonal table. As the CMRIO model ncorporates a huge amount of data ncludng natonal IO data, provncal IO data, customs data, and transportaton data, errors are unavodable. These errors are prmarly from the followng sources: (1) the process of convertng the provncal tables to four-column tables and estmatng the data for the mport-nflow non-compettve tables; (2) the calculaton of control totals of ndustres of provnces; (3) when usng the natonal IO table as the control, the total nputs are calculated by addng together the ntermedate nputs and the prmary nputs (value added), so dscrepances between the ntermedate uses and the fnal uses are unavodable; and (4) orgnal errors n provncal IO tables. After completng all of the steps n the development of the model, all these errors are fnally reflected n the error term. Therefore, t s dffcult to fnd the approprate method to express the balance relatons n the CMRIO models, to calculate control data, and to mnmze the error durng the development of the models. SIC therefore consdered and tested varous methods durng the development of the model and have succeeded n mnmzng the number of errors. As a result, SIC has made only mnor adjustments to the error term and completed development of the model Complaton of the 2005 Transnatonal Interregonal IO Tables Snce the Great East Japan Earthquake of 11 March 2011, the transnatonal economc mpact on specfc domestc regons has attracted a great deal of attenton from regonal governments. Gven the deepenng economc nterdependence and ncreasng transnatonal spllover mpact at the domestc regonal level among East-Asan economes, IDE launched n 2012 a new nternatonal project utlzng the 2005 TIIO table for three East Asan countres: Chna, Japan, and South Korea. We are pleased to report that the 2005 nterregonal IO tables for Japan and South Korea are now avalable and that the SIC of Chna s currently complng the 2002 and 2007 nterregonal IO tables, whch has been offcally publshed (see Zhang and Q, 2012). We also expect to collaborate wth the offcal nsttutons n charge of constructng natonal nterregonal IO tables of these countres: SIC of Chna, the Bank of Korea, and the Mnstry of Economy, Trade and Industry (METI) of Japan. Ths collaboraton scheme helps to dstngush the product among MRIO tables by grantng an offcal property to the data. Acknowledgements The authors are very grateful to Erk Detzenbacher for revewng the manuscrpt, he made very mportant suggestons whch enhanced ths paper. We also thank the other three

17 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 137 anonymous revewers for ther valuable comments and suggestons to mprove the qualty of the paper. References Baldwn, R. (2011) Trade and Industralsaton after Globalsatons 2nd Unbundlng: How Buldng and Jonng a Supply Chan Are Dfferent and Why It Matters, NBER Workng Paper Seres, 17716, Natonal Bureau of Economc Research, Cambrdge, Massachusetts, USA. Inomata, S. and H. Kuwamor (2008) Complaton and Use of the 2005 Asan Internatonal Input Output Table, Asan Internatonal Input Output seres, No. 70, IDE-JETRO. Inomata, S. and B. Meng (2011) Cross-regonal Spllover of Economc Growth: The Terrtoral Impact of Global Manufacturng n Chna. In: WTO-IDE (eds.) Trade Patterns and Global Value Chans n East Asa: From Trade n Goods to Trade n Tasks. Geneva, WTO Secretarat, Insttute of Developng Economes (IDE) (2003) JETRO, Mult-regonal Input Output Model for Chna 2000, IDE Statstcal Data Seres, No. 86, Insttute of Developng Economes JETRO, Chba, Japan. Insttute of Developng Economes (IDE) (2006) JETRO, Asan Internatonal Input Output Table 2000: Volume 1 Explanatory Notes, Statstcal Data Seres, No. 89, Insttute of Developng Economes JETRO, Chba, Japan. Insttute of Developng Economes (IDE) (2007) JETRO, Transnatnal Interregonal Input Output Table between Chna and Japan 2000, Asan Internatonal Input Output Seres, No. 68, Insttute of Developng Economes JETRO, Chba, Japan. JSE (1983) Dstance Tables for World Shppng, 8th Ed., Tokyo, Japan, Japan Shppng Exchange, Inc., Tokyo, Japan. Kuwamor, K. (2006) The Role of Dstance n Determnng Internatonal Transport Costs: Evdence from Phlppne Import Data, IDE Dscusson Paper Seres, No. 60, Insttute of Developng Economes JETRO, Chba, Japan. Meng, B., Y. Zhang, J. Guo, and Y. Fang (2012) Chna s Regonal Economes and Value Chans: An Interregonal Input Output Analyss, IDE Dscusson Paper Seres, No. 359, Insttute of Developng Economes JETRO, Chba, Japan. OECD-WTO (2012) Trade n Value-Added: Concepts, Methodologes and Challenges (Jont OECD-WTO Note). Pars, France, OECD. Shbasak, R., K. Yonemoto, and T. Watanabe (2008) On the Effects of Trade Lberalzaton Polces on Regonal Economes Based on Transnatonal Interregonal Input Output Table Between Chna and Japan, Conference paper presented at the 11th Annual Conference on Global Economc Analyss, Helsnk, Fnland. Watanabe, T. (1966) Internatonal Input Output Table: Structure and Possbltes, IDE Research Documents, No 92 (n Japanese), Insttute of Developng Economes, Tokyo, Japan. WTO-IDE (2011) Trade Patterns and Global Value Chans n East Asa: From Trade n Goods to Trade n Tasks, WTO-IDE Jont Report, Geneva, WTO Secretarat. Zhang, Y.X. and S.C. Q (2012) Chna Mult-regonal Input Output Models for 2002 and Bejng, Chna. Chna Statstcs Press. Zhang, Y.X., Y. Lu, and P. An (2012) The Contrbutons of Chna Japan Economc Relatonshp to Employment and Labor Compensaton Based on 2007 Chna Japan Blateral Input output Table, Paper for Workshop on Chna and Japan Internatonal IO Table.

18 APPENDIX 1. LAYOUT OF AIIO TABLE code Indonesa Malaysa Phlppnes Intermedate Demand (A) Fnal Demand (F) Export (L) Sngapore Thaland Chna Tawan Korea Japan USA Indonesa Malaysa Phlppnes (AI) (AM) (AP) (AS) (AT) (AC) (AN) (AK) (AJ) (AU) (FI) (FM) (FP) (FS) (FT) (FC) (FN) (FK) (FJ) (FU) (LG) (LH) (LO) (LW) (QX) (XX) Indonesa (AI) A II A IM A IP A IS A IT A IC A IN A IK A IJ A IU F II F IM F IP F IS F IT F IC F IN F IK F IJ F IU L IG L IH L IO L IW Q I X I Malaysa (AM) A MI A MM A MP A MS A MT A MC A MN A MK A MJ A MU F MI F MM F MP F MS F MT F MC F MN F MK F MJ F MU L MG L MH L MO L MW Q M X M Phlppnes (AP) A PI A PM A PP A PS A PT A PC A PN A PK A PJ A PU F PI F PM F PP F PS F PT F PC F PN F PK F PJ F PU L PG L PH L PO L PW Q P X P Sngapore (AS) A SI A SM A SP A SS A ST A SC A SN A SK A SJ A SU F SI F SM F SP F SS F ST F SC F SN F SK F SJ F SU L SG L SH L SO L SW Q S X S Thaland (AT) A TI A TM A TP A TS A TT A TC A TN A TK A TJ A TU F TI F TM F TP F TS F TT F TC F TN F TK F TJ F TU L TG L TH L TO L TW Q T X T Chna (AC) A CI A CM A CP A CS A CT A CC A CN A CK A CJ A CU F CI F CM F CP F CS F CT F CC F CN F CK F CJ F CU L CG L CH L CO L CW Q C X C Tawan (AN) A NI A NM A NP A NS A NT A NC A NN A NK A NJ A NU F NI F NM F NP F NS F NT F NC F NN F NK F NJ F NU L NG L NH L NO L NW Q N X N Korea (AK) A KI A KM A KP A KS A KT A KC A KN A KK A KJ A KU F KI F KM F KP F KS F KT F KC F KN F KK F KJ F KU L KG L KH L KO L KW Q K X K Japan (AJ) A JI A JM A JP A JS A JT A JC A JN A JK A JJ A JU F JI F JM F JP F JS F JT F JC F JN F JK F JJ F JU L JG L JH L JO L JW Q J X J USA (AU) A UI A UM A UP A US A UT A UC A UN A UK A UJ A UU F UI F UM F UP F US F UT F UC F UN F UK F UJ F UU L UG L UH L UO L UW Q U X U Freght and Insurance (BF) BA I BA M BA P BA S BA T BA C BA N BA K BA J BA U BF I BF M BF P BF S BF T BF C BF N BF K BF J BF U Import from Inda (CG) A GI A GM A GP A GS A GT A GC A GN A GK A GJ A GU F GI F GM F GP F GS F GT F GC F GN F GK F GJ F GU Import from Hong (CH) A HI A HM A HP A HS A HT A HC A HN A HK A HJ A HU F HI F HM F HP F HS F HT F HC F HN F HK F HJ F HU Kong Import from EU (CO) A OI A OM A OP A OS A OT A OC A ON A OK A OJ A OU F OI F OM F OP F OS F OT F OC F ON F OK F OJ F OU Import from the ROW (CW) A WI A WM A WP A WS A WT A WC A WN A WK A WJ A WU F WI F WM F WP F WS F WT F WC F WN F WK F WJ F WU Dutes and Import (DT) DA I DA M DA P DA S DA T DA C DA N DA K DA J DA U DF I DF M DF P DF S DF T DF C DF N DF K DF J DF U Commodty Taxes Value Added (VV) V I V M V P V S V T V C V N V K V J V U Total Inputs (XX) X I X M X P X S X T X C X N X K X J X U Sngapore Thaland Chna Tawan Korea Japan USA Export to Inda Export to HongKong Export to EU Export to ROW Statstcal Dscrepancy Total Outputs 138 B. MENG et al. Source: IDE (2006). Note: Each cell of A** and F** represents a matrx of and 76 4 dmenson, respectvely.

19 IDE-JETRO S INTERNATIONAL INPUT OUTPUT TABLES 139 APPENDIX 2. ADJUSTMENT OF PRESENTATION FORMAT In , a survey was conducted n order to construct a basc set of nformaton to ad n the desgn of the AIIO common format and adjustment rules. Based on the survey results, we shall brefly look at the areas of conflct where each country s treatment s not n lne. Referrng to Table 1, the man adjustment targets for each natonal IO table can be determned as follows: (1) The Sngapore table should be adjusted from basc prces to producer s prces. Ths s manly due to the fact that most natonal IO tables for Asan countres are valued at producer s prces. In the Sngapore table, all commodty taxes are ncluded n an ndependent row vector of commodty taxes n the value added. Usng ths nformaton and other relable taxaton-related data sources, such as the Publc Fnance Yearbook of Statstcs, we dstrbute the tax to convert the table to producer s prces, at the same tme we keep control totals (output by sector) of the table unchanged. (2) The general prncple concernng prvate consumpton expendture (PCE) n the AIIO table s that the PCE should be valued on a natonal bass, commodty by commodty. However, n Malaysa, Sngapore, and the USA tables, the non-resdent s expendtures n the domestc market are ncluded; n the USA table, the resdent s expendtures abroad s not ncluded. Usng relable statstcal nformaton, we calculate the dstrbuton ratos for foregner s expendtures on home goods and servces and the dstrbuton ratos of resdent s expendtures abroad. These ratos are used to adjust the balance between the PCE and export/mport tems for these three countres. (3) Based on the general prncple of AIIO table, the export vectors should be valued at producer s prces. In Malaysa and Sngapore tables, ths tem s at FOB prces. We use the domestc TTM-related nformaton to make the adjustment for the export tem. At the same tme, the export fgure for transport and trade sectors s changed for keepng the balance. (4) Followng the AIIO prncple, the mport matrx/vector should be valued at CIF prces, not ncludng mport dutes and mport commodty taxes. For Japan, Korea, Phlppnes, Thaland, and the USA, the duty and tax are removed from ther mport fgures by usng the mport dutes and mport commodty taxes-related nformaton. (5) In the AIIO table, no negatve entry should exst n ntermedate transactons, except for those arsng as a result of the generaton of scrap and by-products, and dealngs of second-hand goods. For the Korean table, there are negatve entres along Retal trade (row) where government subsdes are assocated. These negatve values are shfted to subsdes n value-added tems. (6) For the AIIO table, no dummy sector should exst. However, for Chna, there s a standalone scrap and waste sector (row and column), and rural ndustry sector (row and column) n the table; for Japan, there are standalone busness consumptons (n fnal demand and n value added), n-house research, offce supples, and scrap (rows) n the table; for Korea, there are standalone busness consumpton, n-house research, and offce supples n the table; for Malaysa, Sngapore, there s a standalone mputed nterest sector (column only) n the table; and for the USA, there are standalone scrap (row only), used and second-hand goods (row only), royaltes, management of companes, and general government ndustry n the table. All these dummy sectors are