Sectoral wage share and its decomposition in China

ABSTRACT

This paper investigates sectoral contributions to the trend of national wage share, or the labor income share, during 2000–2014 in China. I apply the logarithmic mean Divisia index (LMDI) decomposition, the method widely used in energy studies, to decompose the trend of the wage share. At a sectoral level, with rapid structural transformation, structural change negatively impacted the wage share through the between-sector effect – the structural and price effects – mainly from agriculture. This result confirms Arthur Lewis’s hypothesis that structural transformation has a negative contribution to the wage share. At a national level, when the wage share declined before 2008, the between-sector effect was as significant as the within-sector effect – the wage and productivity effects. After 2008, the within-sector effect directed the increasing wage share trend. This implies that although structural transformation matters to the wage share in a large developing country like China, a wage-productivity nexus has been more influential and determined the increasing trend of the wage share since 2008.

KEYWORDS:

• Labor share of income
• wage share
• decomposition
• real product wage
• developing countries
• China

JEL CLASSIFICATION:

• C43
• D33
• F62
• F63
• O11
• O53

Acknowledgement

A draft version of this paper was virtually presented at the Social Science Institute, the University of Rhode Island, in March 2021 and at the Center for Research on Inequality and Social Policy (CRISP), Thammasat University, Thailand, in August 2021. First, I am grateful to the editor and two anonymous referees for their useful comments. I also want to thank Shanna Pearson, Pengfei Liu, Naphon Phumma, Thanasak Jenmana, and Victor Isidro Luna for all the valuable suggestions and comments. The remaining errors are of course mine.

Disclosure statement

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

Correction Statement

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

Notes

1. Functional income distribution can be separated into the capital share of income (or profit share) and the labor share of income (or wage share). The sum of wage share and profit must always equal 100% (or 1). For example, if the wage share equals 70% (or 0.7), it automatically implies that the profit share is 30% (or 0.3). This paper will use the wage share, the labor share, the labor share of income, and the labor income share interchangeably. All four terms are identical.

²On this point, empirically we can trace it back further to Keynes, 1939 where he points out that the stability of the wage share is “most surprising, yet best-established, facts in the whole range of economic statistics, both for Great Britain and for the United States” (Keynes, 1939, 48).

³The China shock can be marked as the beginning of China’s economic might. The ascendency of China to become one of the economic superpowers was extensively discussed by Arrighi, 2007, Emmott, 2009, Hung, 2009, and Jacques, 2009, among many others.

⁴In fact, the negative economic effect from the 2008 Great Recession in China was minimal due to its unique fiscal stimulus program (Wen and Wu, 2019).

⁵I use percentage rather than the decimal point. The reason is that this unit conforms to the following section where I investigate the change in the wage share, which is very small at the annual level.

⁶All these five countries are categorized as developing economies based on the IMF’s World Economic Outlook 2020 (2020). According to the World Bank’s World Development Indicators, these five countries combined have a labor force equals 1.5 billion in 2014, which is considered 46.72% of the world’s labor force.

⁷This high wage share in agriculture conforms to other literature on China. For example, Zhou, 2015, 154) listed the national wage share along with the wage shares in primary (agriculture), secondary (industry), and tertiary (services) sectors in China during 1993–2004. On average, the wage share in the primary sector was around 85%, whereas the wage shares in secondary and tertiary sectors were 45%-50% and 54%-63%, respectively.

⁸Labor coefficient$\left(Y/L\right)$ is an inverse of labor productivity$\left(Y/L\right)$. The lower labor coefficient automatically implies the higher labor productivity that can be caused by technological progress.

⁹The term ‘stagnant’ does not imply about output growth. Baumol, 2012 proposed that most performing arts or traditional services require the same amount of time and labor as were produced decades ago because new technology cannot easily facilitate its production process. As a result, those services are stagnant because over time the cost per unit of production cannot be reduced as fast as manufacturing.

¹⁰Note that these five countries were at different stages of development in 2000s. Brazil and Mexico were more developed than the three Asian countries. The rise of Latin America in general can be traced back to the inter-war period. For more details of the history of structural transformation and growth in Latin America, see Ocampo, 2003, among others.

¹¹Another possible reason for the rise of manufacturing in China is the new cycle of technology, i.e. the new digital economy (Gordon, 2000). This new cycle should benefit a country like China that focuses on manufacturing exports. I want to thank an anonymous referee on this point.

¹²Since the 2000s, India has adopted a service-led growth strategy rather than the manufacturing-led growth implemented by China. Khurana and Mahajan, 2020 scrutinized data for urban workers in the non-agricultural sector from 1983 to 2017. The country experienced high economic growth and rapid structural changes from agriculture to service sectors during 2004–2012. It dominated world exports in telecommunications and information technology (IT) services. They found that IT and construction sectors showed high employment growth during 2004–2017.

¹³Kim et al., 2019 argued that Indonesia after the 1997 Asian Financial Crisis reoriented its economy toward service. Industrialization from 2000 has been slow coupled with lower productivity growth and stagnate manufacturing employment growth. Overall, the Indonesian economy during the 2000s could not match the spectacular growth numbers generated two decades prior to the 1997 Asian Financial Crisis. This industrialization slowdown may also explain the positive structural effect found in the previous section.

¹⁴Other literature found the rise of the real wages in Brazil only began around 2006. Loureiro and Saad-Filho (2019, p. 70) showed that the index of average earnings from main job in Brazil was stable from 2001–2006. Afterward, the index increased from 103.2 in 2006 to 154.8 in 2014.

¹⁵This estimation is not far-fetched compared to other studies. Rozelle et al., 2020 showed that that between 1995 and 2015, the Chinese economy experienced fast growth in wages in all economic sectors. The fastest growth was found in manufacturing, construction, and the informal service sectors. Athukorala and Wei, 2018 reviewed the Lewisian turning point debate regarding the Chinese labor market. Based on the annual urban household survey, most studies revealed that the real wage had been stagnant until the mid-1990s. Between 1988–2013, however, the real average annual earnings increased by a factor of five, from 3,880 Yuan to 19,674 Yuan (Athukorala and Wei, 2018, 425). Zhang et al., 2011 and Cai and Du (2011), for example, confirmed that the Lewis turning point in China was in 2003, which also confirms my result in the previous section that the wage effect in China significantly increased since 2004.

¹⁶Summa and Serrano, 2018, 365) pointed out that while total productivity increased after 2009, nominal manufacturing productivity declined into negative territory during 2010–2014, which intensified the rise of the wage share in manufacturing and the whole economy.

¹⁷Seguino, 2017 provided a comprehensive review of Amsden’s works regarding the role of wages for late industrializers. Amsden believes that for developing countries, lower wages result in lower productivity. The main reason is that for late industrializers, technology is transferred and learned at the lowest level through “practice, adaptation and experimentation” ⁷Seguino (2017) 103). Higher wages therefore incentivize workers and increase retention rates. Rising wages also pressures firms to innovate. To cut wages, as the Washington Consensus argues for, will result in rising costs and lower labor productivity.

¹⁸In the original version, the Y-axis is labelled$W/L$ . Thus, the iso-curves represent the unit labor cost$\left(W/L\right)$.

¹⁹This modified Amsden model attributes the change in the wage share to the dynamics of real wage and labor productivity. In other words, it ignores the between-sector effect and only focuses on the within-sector effect. I want to thank an anonymous referee for the comments on the model.

²⁰Some may cast doubt if the East Asian approach can be used for a long period of time. Piovani, 2014 argued that the decline of the wage share in China before 2007, which implies a decline in worker’s bargaining power, was inherently unsustainable in the long run due to demographic, social, and environmental reasons.