Is China’s economic growth profit-led or wage-led? A re-estimation incorporating investment nonlinearity, sectoral change, and regional disparity

Abstract

This paper estimates China’s growth regime based on Bhaduri-Marglin model, using the provincial panel data from 1978 to 2017. It contributes to related empirical studies in that: (1) it tests that the responsiveness of investment growth to labor share increase is dependent on the existed labor share level; (2) it uses the labor share changes adjusted for the sectoral-change effects as a more appropriate proxy for distributive changes; and (3) it manifests the necessity of recognizing regional heterogeneity in parameters for a large economy. The results show that at 2017 level, whereas the eastern area is weakly profit-led, the inland area is strongly wage-led, and an even increase of labor share in all areas lead to more rapid growth of the whole economy. At this stage, China needs pro-labor institutional reforms to restore rapid, balanced, and sustainable growth.

Keywords:

• China
• growth regime
• labor share
• wage-led growth
• unbalanced growth

SUBJECT CLASSIFICATION CODES:

• O11
• E12
• E25
• E62

Notes

Disclosure statement

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

Data availability statement

All the data supporting the findings of this study are available within the online supplemental material.

Notes

1 Data from China Statistical Yearbook 2018, http://www.stats.gov.cn/english/Statisticaldata/AnnualData/

2 Nobody has expressed the mainstream consensus more straightforward than Jiwei Lou, then finance minister of China. In his speech made to the school of economics and management of Tsinghua University, on April 24, 2015, Lou bluntly attacked the Labor Contract Law enacted in 2008 and centralized collective bargaining for “it decreases the mobility and flexibility of labor market” as in Europe. Besides, he asserted: “there’s an iron law, that is, under normal conditions, the growth rate of real wage has to be lower than the growth rate of labor productivity.”

3 $\mathrm{dln}D$ refers to instantaneous growth rate of $D$ in differentiable time. It is only an approximation to the growth rate of $D$ in discrete (e.g. annual) time series.

4 The difference specification is a more generalized representation since under cointegration $\frac{\partial \widehat{D}}{\partial \left(\mathit{dLS}\right)}$ and all its lags sum up to $\frac{\partial \mathit{lnD}}{\partial LS}.$

5 Theoretically, this problem can be solved by simply adding sectoral changes as control variables. However, in econometric practices, this solution cannot guarantee a robust estimation of the coefficients of labor share change given the high correlation (i.e. collinearity) between labor share change and sectoral changes, especially for the nonlinear investment function I will present below.

6 To illustrate this, given $\mathrm{d}SW2+\mathrm{d}SW3=-\mathrm{d}SW1,$ Equation (4) is transformed to: $$\mathrm{d}LS=\left(LS2-LS1\right)\mathrm{d}SW2+\left(LS3-LS1\right)\mathrm{d}SW3+\mathrm{d}\mathit{ALS}$$

7 For example, in a function with $\widehat{D}$ as the dependent variable, if the coefficient of $\mathrm{d}\mathit{ALS}$ is $\alpha$ and the coefficient of $\widehat{D}(-1)$ is $\lambda ,$ the long term effect of $\mathrm{d}\mathit{ALS}$ on $\widehat{D}$ is $\alpha /(1-\lambda ).$

8 LSDVC method only allows for first lag of dependent variable as regressor. But this is not a severe limitation for annual data.

9 ${\alpha }_C$ is individual-specific since $LS$ is individual-specific. The area-level ${\alpha }_C$ is derived by aggregating the individuals’ ${\alpha }_C$ weighted by their respective share of $C$ of the area.

10 Although the price effect on import is postulated to be zero, demand effect on import ${\beta }_M$ which is relevant to calculate the multiplier is derived by cointegrated estimation of import function excluding the price variables, since cointegration between $\ln M$ and $\ln Y$ is robust. Estimated ${\beta }_M$ for the eastern and inland areas are 1.252 and 1.305, respectively.

11 This two-step estimation is akin to a 2SLS estimation of $\widehat{Pd}$ with $\widehat{\mathit{nwr}}$ as the instrument of $\mathrm{d}\mathit{ALS}.$ However, I did not run such a 2SLS estimation because: (1) the inclusion of lag of dependent variable will cause the biased estimates, and (2) $\widehat{\mathit{nwr}}$ is not a qualified instrument because apart from influencing the dependent variable through the instrumented variable, it also influences $\widehat{Pd}$ directly.

12 Most related literature seem to follow this assumption, although they do not mention this issue explicitly. An alternative assumption is that government consumption and inventories keep a constant proportion of $Y,$ which will only change the expression of $\beta ,$ without altering the results substantially: $\beta =\frac{C}{Y}{\beta }_C+\frac{I}{Y}{\beta }_I+\frac{X}{Y}{\beta }_X-\frac{M}{Y}{\beta }_M+\frac{\mathit{RES}}{Y},$ where $\mathit{RES}$ denotes the sum of government consumption and inventories.

13 ALSV 1 of the eastern area reached its peak of 4.94 in 2015. Detailed time series of demand structure and ALSV are also accessible in the supplementary material, Part D.

14 Admittedly, any estimation or simulation of the growth potential of increasing labor share should be cautious about the Lucas (1976) critique, i.e., the parametric instability in the event of policy intervention. However, I believe this only matters to the specific magnitude of the growth effect of distributive changes, instead of refuting the nature of the wage-led regime at present and the appropriateness of pro-labor institutions in a foreseeable future.

Additional information

Funding

This research is supported by the National Office for Philosophy and Social Sciences under grants from the National Social Science Fund of China [grant number 16CJL036].

Notes on contributors

Dun Liu

Dun Liu is at School of Economics and Management, Beijing Jiaotong University, Beijing, China. Dun Liu received his Ph.D. in economics in 2013 in Beijing Jiaotong University. Now he is a lecturer in the School of Economics and Management, Beijing Jiaotong University. His research interest focuses on heterodoxy theories on income distribution and economic growth. E-mail: [email protected]