Has the rise of China’s domestic supply chain contributed to its GDP increases?

ABSTRACT

This paper applies a structural decomposition of the input-output tables to investigate the sources of changes in the GDP of China, paying special attention to the import substitution of intermediate inputs. It is argued that the specific way GDP is decomposed in this paper sheds more light on the origin of growth than does the conventional decomposition of GDP by its final demand components. The results of the decomposition show that import substitution of intermediate products, which is an act of de-globalization of trade and a model of industrialization some Asian newly industrializing economies followed decades earlier, also became important in China, but only belatedly: in 2011–2012 and 2013–2014. The phenomenon is found to be significant in the manufacture of (i) computer, electronic and optical products, (ii) chemicals and chemical products, (iii) coke and refined petroleum products, and (iv) basic metals.

KEYWORDS:

• Input-Output
• import substitution
• supply chain
• structural decomposition
• de-globalization

JEL CLASSIFICATION:

• D57
• O10

Disclosure statement

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

Notes

¹ See for example Fei, Ohkawa, and Ranis (1985, 35-64). When South Korea and Taiwan entered the period of import substitution of intermediate goods, Japan was worried that these late-comers could catch up with it. In Japan in the late 1970s, for example, newspapers “were filled with wary editorials about the ‘Korean challenge’” (Cumings 1984, 36). It should be noted however that the division into such distinctive stages of industrialization points more to the broad direction than to the actual working of individual sectors or products. In reality, import substitution and export expansion often co-exist in any sector consisting of individual commodities reflecting different levels of technology. The general direction has been that late-comers would try to “move upstream or upwards” by replacing intermediate products or equipment that were formerly out of their reach and had to be imported with domestically produced ones.

² Based on an analysis of disaggregated trade data, Cui and Syed (2007) find that China’s exports have started to rely more on domestically sourced components. Shin and Choi (2016, 17) find that “China is shifting away from assembling imported parts and components to domestic sourcing. Len (2017) cites a report which anticipates a possible squeezing out of Taiwan-based information technology (IT) supplies by China’s local firms.

³ But the authors find that the sectors likely to be labelled as relatively sophisticated such as electronic devices have particularly low domestic content (about 30% or less).

⁴ See also Yang and Yang (2017).

⁵ In the case of China, see, e.g. Liu (1998); Andreosso-O’Callaghan and Yue (2002); and Pei, Oosterhaven, and Dietzenbacher (2012).

⁶ Here it means fixed capital formation and changes in inventory, with the latter being typically very small in quantity compared to the former.

⁷ The superscript d is missing on e, as presumably all exports are domestically produced.

⁸ Strictly speaking, private consumption should be a function of disposable income, which is income minus taxes. The current equation does not deal with taxes, so the “savings rate” here incorporates the concept of income taxes, and one minus the savings rate is simply the percentage of private consumption in total income (GDP).

⁹ Government consumption of domestic products is by itself assumed to be exogenous, so there is no further decomposition parallel to the treatment of private consumption.

¹⁰ The variables other than the one for which a difference is taken are evaluated at their average values between the two (polar) periods. This is a good approximation of the average of all possible forms of decomposition. More on this later.

¹¹ In the current formulation, ∆[(1 – a^d/ d)/(1 – a^d)] is decomposed first into those caused by ∆d (the first term of the RHS of this equation) assuming a^d is constant, and those caused by ∆a^d assuming d is constant. The latter part is then further decomposed into the second and the third term in the RHS of Equation (10). While interpreting the meanings of the second and third terms, one should bear in mind that d is assumed constant, so changes in a^d (the “domestic input coefficient”) mean exactly the same as changes in a (=a^d/ d, the “input coefficient”). For this reason, the second and third terms can alternatively be called the “input coefficient effect” and the “input coefficient induced Leontief inverse effect”, respectively. Also for this reason, in the following text, when the effect embodied in the second term is alluded to, it is sometimes referred to as the “(domestic) input coefficient effect”.

¹² In the single-sector model, there is of course no Leontief inverse “matrix”, but here the analogy is used as that inverse is precisely (1 – a^d)^–1 given a single sector.

¹³ The d = a^d/a in Equation (9) can be alternatively expressed as a = a^d/d.

¹⁴ Various possible forms of decompositions exist depending on the arrangement of beginning and end points. Dietzenbacher and Los (1998) analyse empirically the sensitivity of the results to the various formulations, and found that the average of the polar decompositions is a good approximation of the mean of all possible forms. This option is the one adopted here.

¹⁵ This paper uses the terms “intermediate inputs”, ‘intermediate products’ and “intermediate goods” interchangeably.

¹⁶ This 2016 release of the WIOD has 56 sectors. The values across sectors are summed up to obtain the aggregated figures in this single-sector model. For an explanation of the construction of cross-country input-output tables, see Dietzenbacher et al. (2013).

¹⁷ To be more accurate, the consumers saved more or were taxed more by the government or both.

¹⁸ This is consistent with the findings of Sasaki and Satoko (2009).

¹⁹ The mutually offsetting effects of the domestic input coefficient in column (1) and the Leontief inverse in column (3) are not included in the ranking.

²⁰ Figures of the top three contributing sectors are highlighted in Table 2.

²¹ It is to be noted, however, that there are several sectors with sizable negative contributions.

²² As mentioned above, there has been no significant import substitution with regard to China’s gross capital formation or investment in the years covered by this paper.

²³ For a discussion of the sophistication of China’s exports compared to other countries, see Kwan (2002). For assessing the degree of sophistication of exports and its implications for economic growth, see Hausmann, Hwang, and Rodrik (2007). For the application of such a concept to China, see, e.g. Xu (2010) and Yu and Hu (2015).

²⁴ See the discussion of the “flying geese paradigm” in Cumings (1984, 1–4), and the product cycles theory proposed by Vernon (1996). For a discussion of the implications for China, see, e.g. Widodo (2008) and Nakagane (2013). See also Wan (2003).

²⁵ See for example Yao (2011).

²⁶ According to Perkins and Rawski (2008) and Perkins (2018), in the high-growth post-reform period of 1978–2005, China’s total factor productivity (TFP) growth reached an average of 3.8% per annum, accounting for about of 40% of GDP growth. After 2005, the increase in TFP fell from an average of 3% per annum in 2006–2011 to 1% per annum in 2012–2016. Such results are broadly consistent with those obtained by other researchers.

²⁷ In the sense of cross-border direct investment, this would be an act of more globalization.

²⁸ Giammetti et al. (2022) use the EUREGIO database that includes data for 14 industries in 246 NUTS 2 regions of the EU-25, plus data at the country-level for the same 14 industries in Bulgaria, Romania, and other 14 extra-EU trading partner countries. They investigate the effects of three scenarios of deglobalization: (1) the end of foreign intermediate input flows; (2) a return to the 2000 production schemes and trade patterns; (3) the Europeanization of the global value chains.

²⁹ They describe their approach as an application of the global extraction method (GEM), under which when an industry in an economy is extracted, they have to specify the extracted amounts are replaced by which economies and by how much respectively. They apply their model to the hypothetical extraction of the motor vehicle industry in China, the US, and Germany, using the 2014 WIOD input – output table.