![Sample our Environment & Sustainability journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5935/TOC-Subject-Sample-2021-Environment-Sustainability.jpg"})
Abstract
This paper examines Japan’s participation in global value chains (GVCs). To this end, we use plant-level data for Japan to split output in each industry in Japan’s manufacturing sector into output for export or domestic sale and create an extended multi-country input–output table (MIOT). We then compute trade in value added (TiVA) indicators to examine the participation of Japanese manufacturing plants in GVCs. Our estimates suggest that Japan’s forward participation in GVCs is lower than suggested by estimates computed from a traditional MIOT. We infer that this result is due to high cross-border production fragmentation as well as the large presence of Japanese multinational companies in global manufacturing and the high volume of intra-firm trade in Japan’s manufacturing sector. We conclude that considering firm heterogeneity in production for export and domestic sale in MIOTs provides a more accurate understanding of global production fragmentation.
Acknowledgements
This study was supported by the Research Institute of Economy, Trade and Industry (RIETI). The authors would like to thank Makoto Yano, Masayuki Morikawa, and other seminar participants at RIETI for their constructive comments and suggestions. This paper was presented at the Second Meeting of the OECD Expert Group on Extended Supply-Use Tables in Paris, France, the meetings of the Working Party on Industry Analysis (WPIA) in Paris, France, and the 24th IIOA conference in Seoul, Korea. The authors would like to thank meeting and conferences participants for their valuable feedback. The authors also would like to thank Ralph Paprzycki for the professional proofreading of the manuscript. Finally, comments from the editor and two anonymous referees have helped to shape the structure of this paper to more clearly bring out our key argument. The authors used the micro-data of the 2012 Economic Census for Business Activity (Ministry of Internal Affairs and Communications and Ministry of Economy, Trade and Industry) and the 2012 Basic Survey on Wage Structure (Ministry of Health, Labour and Welfare). Regarding the use of the data, the authors are grateful for the help of the ministries and the Quantitative Analysis and Database Group of RIETI. Koji Ito would like to thank John Gilbert for his helpful suggestions on GAMS programing. All remaining errors are our own.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In this study, we distinguish between sectors and industries. The term ‘sector’ refers to broad classifications such as the manufacturing sector and the service sector. We reserve the term ‘industry’ for categories within these sectors, such as the textile industry in the manufacturing sector.
2 Note, that TiVA indicators do not include factor inputs induced by foreign final demand. We compute FFD_F from both the original non-split and the extended split version of the ICIO table and compare the results obtained. We use matched employer-employee data to obtain factor inputs for the split ICIO table.
3 Piacentini and Fortanier (Citation2015) also compare VA/turnover ratios, but the results are mixed.
4 Note that we have information on the number of workers from the ECBA micro-data, so that we are able to split the ICIO table by plant size following our methodology. We are planning to extend our work along these lines to confirm the inferences of our analysis.
5 We thank an anonymous referee for pointing out that Chen et al.’s (Citation2018) study is similar to ours in spirit, since they distinguish production for processing exports, domestic use, and non-processing exports for China and create an extended MIOT. We believe that distinguishing production for export and domestic sale using plant-level data to split Japan’s manufacturing sector in the OECD ICIO table is a novel approach to create an extended MIOT that considers firm heterogeneity. The nature of production for export differs from the nature of production for processing exports because the former implies the existence of special economic zones that alter firms’ incentives and behaviour.
6 In our IO analysis we assume that service industries use the same technology for production for export and for domestic sale.
7 Land is not included in the value of tangible fixed assets.
8 We work with the ICIO SNA93, ISIC REV.3 version of the OECD ICIO table.
9 For the factor input analysis we use the employer–employee matched data to compute labor input as described in Section 6.
10 Information on imported input intensities for exporting and non-exporting plants is not available in our data. The dataset contains information on plants’ total amount of purchased intermediate goods but not on the sector and country from which these intermediate goods are purchased.
11 In JSIC rev.11, there are 637 industries, which are represented by three-digit codes.
12 The ICIO table also contains data on domestic sales and exports at the industry level, and we could split the table using this data. However, we use the information from the ECBA, because the ECBA is an almost complete census and hence provides more accurate information.
13 Note, that the labelling that follows (i.e., a, b, c, d, e, f, and g) corresponds to the labelling in Figure (b).
14 This calculation is based on the assumption that technology is homogenous within plants producing for export and domestic sale.
15 We do not use information on value added from the micro-level data. That would mean making additional restrictive assumptions on the distribution of value added.
16 The purpose of this estimation is twofold. The first is to ensure that the balance conditions in the aggregated ICIO table are always satisfied, and that the estimated ICIO table is consistent with the original ICIO table. The second is to ensure that the estimated split ICIO table is consistent with the structure of production for export and domestic sale. Our estimation framework closely follows Ma et al. (Citation2015). Details of the estimation framework are in Appendix 1.
17 The part of the interindustry transaction matrix Z and the output vector y that corresponds to Japan’s manufacturing sector is split into production for export and domestic sale, as explained in Section 3.2. See Appendix 1 for technical details.
18 We use version 2 of the definitions of TiVA 2015 indicators. Retrieved from https://www.oecd.org/sti/ind/tiva/TIVA_2015_Indicators_Definitions.pdf on June 19, 2017. During the writing of this manuscript, a newer version has been released by the OECD, which can be found at http://www.oecd.org/sti/ind/tiva/TIVA_2016_Definitions.pdf. The calculations remain the same.
19 See Appendix Table 1.
20 Note that DVA is higher for the coke and petroleum industry. This industry is somewhat exceptional. In total, the number of plants is smaller in this industry and the share of exporters is relatively high (around 30%). Higher labor productivity of exporters could explain this result of a higher DVA.
21 Exceptions are the rubber and plastics products and the coke, refined petroleum products, and nuclear fuel industries.
22 Detailed results are provided in Appendix Table 1.
23 The ECBA and BSWS use identical identification numbers for prefectures, cities, and plants, allowing us to merge the ECBA and BSWS data.
24 The merged data cover 93.8% of the employee data and 94.0% of the establishment data in the original BSWS dataset.
25 Detailed data and information are available upon request.
26 The approach used to calculate the factor content is explained in Appendix 2.
27 Detailed data are available upon request.
28 We also examined the difference between factor inputs embodied in global final demand calculated from the split and non-split ICIO tables. We found that using a non-split IO table overestimates the level of factor inputs induced by foreign final demand. The details are available upon request.