Technology level and the global value chain

ABSTRACT

This paper investigates the role of technology levels in shaping the structure of the global value chain (GVC) at the macro level. We incorporate the Ricardian comparative advantage into the production of intermediate goods involving both snake-and spider-type supply chains to capture the overall GVC integration. We analytically find that the country with a higher technology level produces the intermediate inputs at production stages involving a higher degree of complexity, which is consistent with the real data. Finally, we calibrate the GVC participation value and wages and document the good fit of our model to the real-world data.

KEYWORDS:

• Global value chain
• technology level
• participation value
• World Input-Output Database

JEL CLASSIFICATION:

• F12
• F15
• O33

Acknowledgment

We would like to thank Alan V. Deardorff, Artem Kochnev, Etsuro Shioji, Francois Gabe, John P. Tang, Jota Ishikawa, Kalina Manova, Konstantin Kucheryavyy, Kyoji Fukao, Minoru Ichikawa, Naohito Abe, and Pol Antràs for their insightful discussions. Our special thanks go to the editor, an anonymous referee, Hiroya Akiba, Hayato Kato, Hiroshi Morita, Kiyoyasu Tanaka for their careful reading of the previous versions of this paper, and for helpful comments that assisted us in considerably improving this work. In addition, we are grateful to the seminar and conference participants at the XXI April International Academic Conference on Economic and Social Development, the 72 European Meeting of the Econometric Society, the 14 Australasian Trade Workshop, the 15 International Conference (Western Economic Association International), the 11 FIW–Research Conference in International Economics, the 2018 Spring Meeting of Japanese Economic Association, the 2018 Asian Meeting of the Econometric Society, the Dynamics, Economic Growth and International Trade DEGIT XXI Conference, the 74 Annual Meeting of the Japan Society of International Economics, and Hitotsubashi University for their useful comments. Financial support was gratefully received from the Japan Society for the Promotion of Science (Grant Number: 18K12762), the Joint Usage and Research Center, Institute of Economic Research, Hitotsubashi University (Grant ID: IERPK2212) and the Hirose International Scholarship Foundation. Support from the Faculty of World Economy and International Affairs at HSE University is gratefully acknowledged. All remaining errors are our own responsibility.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

¹ Chor (2019) provides a review of theoretical modelling of the GVC, and Johnson (2018) summarizes the key empirical results.

² The terms snake and spider are from Baldwin and Venables (2013). In a snake production process, parts need to be performed sequentially, whereas in a spider production process, separate parts are assembled in no particular order.

³ We offer an empirical evidence of this proposition in Appendix A.

⁴ The WIOD release, 2016, follows the ISIC rev. 4 classification. There are 19 manufacturing industries within 56 sectors in the WIOD.

⁵ Unfortunately, the available data do not allow us to make a weaker assumption. It is true that the output of a stage in our theoretical model is one intermediate good, whereas a sector in WIOT produces more than one intermediate good as well as final goods.

⁶ In what follows, we omit the time superscript to avoid notational ambiguity.

⁷ This approach is similar to the market concentration definition of firms in the industry. We used alternative values of 50%, 60%, and 80% and did not observe any significant changes in the results. Hereafter, we refer to 60% and 80% thresholds when reporting the results. The full results are available upon request.

⁸ With this $T_{max}$, the positivity of the argument of the logarithm in Equation (7)(7) $y{\int }_{j_{K-1}^{\ast }}^{T_{max}}{a}_{j,K}dj=L_K\iff yln\frac{T_K-j_{K-1}^{\ast }}{T_K-T_{max}}=L_K.$(7) is guaranteed.

⁹ This is a technique for solving $n$ equations of a system one at a time in sequence, and it uses previously computed results as a new guess for the next iteration round as soon as they are available. For details, see Judd (1998).

¹⁰ To determine real compensation, we deflate nominal compensation using the GDP deflator. GDP deflator data are obtained from the World Bank World Development Indicators. For Taiwan, the GDP deflator is obtained from the Statistics Bureau of Taiwan (https://eng.stat.gov.tw). Note that the definition of the wage in Equation (12)(12) $w_k=\frac{\sum _j{w}_{j,k}}{\sum _j{L}_{j,k}}.$(12) is identical to country-level wages computed as a weighted average of industry wages using as weights the weight of the industry labour force in the total labour force, by country.

¹¹ We follow the standard approach for input-output analysis Miller and Blair (2009). The calculation of value added in world final demand is discussed in OECD (2019).

¹² The ${44}^{th}$ column vector corresponds to the rest of the world.

¹³ Participation values are deflated using the value added price index from Socio-economic Accounts.

¹⁴ Note, that the input data for calibration are technology level and employment. To calculate these data, we rely on value added from the WIOT and employment derived from Socio-economic Accounts of the WIOD. The counterpart of observed data of the GVC participation is computed using input-output linkages in the WIOT. Employees compensation represents a distinct vector in Socio-economic Accounts of the WIOD. Therefore, we do not expect that there is a correlation driven by circular use of data.

¹⁵ For details, see Section 3.1.

¹⁶ Similar to the previous proof, if we consider any two arbitrary countries, North ($N$) and South ($S$), where the North has a higher level of technology than the South, and two arbitrary production stages, $i$ and $j$, where stage $i$ is more complex than stage $j$, we always can prove that $\frac{a_{i,N}}{a_{j,N}}<\frac{a_{i,S}}{a_{j,S}}$.

¹⁷ See Appendix B.

¹⁸ The reported shares from Statista are for 2018. https://www.statista.com/statistics/1106909/china-global-production-share-by-industry/ accessed on September 20, 2022.

¹⁹ We used the weighted distance from the CEPII GeoDist database (Mayer and Zignago 2011). The internal distance in China is a population-weighted distance between major cities.

²⁰ The most common assumption for the trade cost function is $tradecost=distanc{e}^{\rho }$, where $\rho$ is estimated to be around 0.3 (Anderson and Van Wincoop 2004).

Additional information

Funding

Financial support was received from the Japan Society for the Promotion of Science (Grant Number: 18K12762), the Faculty of World Economy and International Affairs at HSE University, the Joint Usage and Research Center, Institute of Economic Research, Hitotsubashi University (Grant ID: IERPK2212) and the Hirose International Scholarship Foundation.