本文已发表于《The World Economy》2017年第11期，第2378-2402页。
作者：Xun Zhang, Guanghua Wan, Chen Wang, Zhi Luo（罗知）
Introduction: There is a sizable and growing literature, focusing on the determinants of income inequality (e.g. Gottschalk & Smeeding, 2000; Greenwood, Guner, Kocharkov, & Santos, 2014; Lerman & Yitz-haki, 1985; Li, Squire, & Zou, 1998; Piketty & Saez, 2003). And more and more research atten-tion has been directed towards the role of technical change as a major driver of income distribution (Acemoglu, 1998).
The conventional approach to analysing the technology–inequality nexus is to identify and esti-mate the impacts of technical change on the wage gap between skilled and unskilled labour, typi-cally in terms of the difference in the average income between these two groups of labourers. According to Acemoglu (1998) and Katz and Murphy (1992), new technologies lead to increases in the productivity of skilled workers and their wages, enlarging this wage gap. Krusell, Ohanian, Rıos-Rull, and Violante (2000) argue that improvement in capital-embodied productivity leads to rising demand for equipment and, when equipment is complementary with skilled labour, the wage gap rises. This gap-enlarging finding has been confirmed by many scholars, including Aghion, Howitt, and Violante (2002), Esquivel and Rodrıguez-Lopez (2003), Moore and Ranjan (2005), and Van Reenen (2011). On the contrary, Goldin and Katz (1996) found that this gap was kept in check in the USA despite significant technological progress. Card and DiNardo (2002) concluded that wage inequality measured as the standard deviation of log wages and the 90th and 10th per-centile wage gap was stable in the 1990s in the USA despite advances in computer technology.
However, wage inequality, especially the wage gap between skilled and unskilled labours, is only one component of the overall inequality, notwithstanding its importance. By definition, total inequal-ity can be expressed as a weighted sum of labour income and capital income (CI) inequalities.1 On the other hand, a driver of income distribution such as technical change may generate different impacts on the overall inequality than its components. For example, an anti-discrimination policy may help narrow the gender gap but may lead to higher wage inequality within male employees at the same time. Similarly, capital-augmenting technical change may enlarge the wage gap between the skilled and unskilled but could meanwhile help reduce inequality within the capitalists, leaving its overall impact on the overall income inequality underdetermined. Clearly, it is insufficient to just analyse the technical change–wage gap nexus if one is interested in the overall income inequality.
To the best of our knowledge, little has been published on the technical change–income inequal-ity relationship, with the exception of Jaumotte, Lall, and Papageorgiou (2013) who found a positive impact of technological progress (defined as the share of ICT capital in total capital stock) on income inequality based on a panel data set of 51 counties over a 23-year period from 1981 to 2003.
This paper represents an early attempt to gauge the impact of technical change on the overall inequality, not just a particular component of inequality. This is achieved by establishing that the labour share of income is negatively correlated with overall inequality as indicated by the popular Gini coefficient, and by modelling the labour share of income as a function of technical change. Based on 1978–2012 provincial panel data from China, the framework of Acemoglu (2002, 2007) will be employed to measure technical change. And the labour share of income will be then regressed on the estimated technical change. The main empirical results show that technical change in China had been mostly capital-biased. It contributed to the successive reductions in China’s labour share of income and thus rapid rises in income inequality.
The rest of the paper is organised as follows. Section 2 presents our analytical frameworks, including arguments for establishing the correlation between the labour share of income and tech-nical change and that for measuring technical change. In Section 3, we discuss data and empirical econometric models. Section 4 provides estimation results and discussions. Finally, Section 5 con-cludes.