Revisit the Nexus between Saving and Inequality in Labor Intensive Economies: Evidence from China

ABSTRACT

Using an extended overlapping generations (OLG) model, we theoretically prove that functional inequality resulting from weak labor bargaining power can be a key driver of high saving rates, as observed in China and other labor-abundant Asian emerging markets. Income distribution that favors capital over labor may attract excess capital investments and hence lead to high saving rates. The link between inequality and saving is especially prominent for the household sector because excess return on capital motivates the working-age population to increase their retirement savings. We also find empirical support for our theoretical predictions using China’s sectoral-level data.

KEYWORDS:

• Saving rate
• income inequality
• OLG model
• emerging markets
• labor bargaining power

JEL CLASSIFICATION:

• D30
• D91
• E21

Author agreement

Both authors read and approved the final manuscript. The article is the authors’ original work, hasn’t received prior publication and isn’t under consideration for publication elsewhere.

Authors’ contributions

XC is the main contributor. She set up the theoretical framework and has contributed to the empirical analysis and paper writing. JL has contributed to paper writing, paper revisions and manuscript preparation. ZL has redesigned the empirical framework, and contributed to empirical analysis and paper revision in the resubmission stage.

Authors’ information

Both Dr. Xiao Chang and Dr. Jingya Li are the assistant professor of the Division of Business and Management, BNU-HKBU United International College. Dr. Zongyuan Li is the assistant professor of the Department of Financial Engineering, School of Maritime Economics and Management, Dalian Maritime University.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/1540496X.2022.2084378

Disclosure statement

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

Notes

1. Even though sectoral income distribution has not been extensively discussed in the literature, it is important in China (Yang, Zhang, and Zhou 2012) and can affect sectoral saving rates (Chang et al. 2020). As exhibited in the left panel of Figure 1, sectoral inequality (distributing income in favor of non-household sectors) has deteriorated since the late 1990s.

2. A possible reasons is the complex social security system, which includes both progressive and regressive social security policies (Cai and Yue 2020).

3. Since $Y_1=wL=\eta Y$, we know $w=\eta k^{\alpha }$ where $k=K/L$. From $M{P}_L=\left(1-\alpha \right)k^{\alpha }$ and $\eta <1-\alpha$, it follows that $w<M{P}_L$ or $w+{\sigma }_w=M{P}_L$ where ${\sigma }_w>0$. Similarly, since $r=\left(1-\eta \right)k^{\alpha -1}$ and $1-\eta >\alpha$, we see that $r>M{P}_K$ or $r-{\sigma }_r=M{P}_K$ where ${\sigma }_r>0$. Clearly, a lower η increases ${\sigma }_w$ and ${\sigma }_r$ given k, since ${\sigma }_w=\left(1-\alpha -\eta \right)k^{\alpha }$ and ${\sigma }_r=\left(1-\alpha -\eta \right)/k^{1-\alpha }$ where $\eta +\alpha <1$.

4. Our OLG modeling is kept at a minimal level of complexity since it is aimed only at showing the effect of inequality on saving. One may proceed with a deeper exploration in future research by including: (a) inequality between hand-to-mouth workers and wealthy individuals, (b) social security for the elderly, and (c) bequests left to children. An OLG framework can accommodate heterogeneous utility maximizing agents, social security programs, and one- or two-sided altruism, but its dynamic analysis, if using a new rate of total savings (as defined shortly in this paper), would become much more complicated than traditional OLG models.

5. It is interesting to examine a three-generation OLG model in future research by explicitly considering the cohort of children. One can then study the impact of changes in youth dependency on the saving rate. However, we will not do so here to conserve space.

6. Because youth dependency ratio and life expectancy are highly correlated (with a correlation coefficient of 0.972), adding both to the empirical model will lead to the problem of multicollinearity.

Additional information

Funding

This paper was research supported by Higher Education Teaching Research and Reform Project in Guangdong Province [under grant code UICR0400014-21] and BNU-HKBU United International College [under grant code R72021117].