Investment in Infrastructure and Regional Growth in China

ABSTRACT

In this paper we empirically examine the possible effects of regional infrastructure investment on regional per capita income growth and regional total factor productivity (TFP) growth. We first present a theoretical framework for output decomposition and then design a panel data growth model to examine the possible impact of regional infrastructure investment on regional per capita income growth. We find a negative effect of the former on the latter. We also perform a variance decomposition exercise and examine the possible effect of regional infrastructure investment on regional TFP growth. Our results suggest that regional infrastructure investment does not affect regional economic growth through the former’s impact on regional TFP growth. We also examine the possible relationship between regional infrastructure investment and the estimated individual region effects and find a negative correlation between the two.

KEY WORDS:

• growth
• inequality
• Infrastructure
• total factor productivity

JEL:

• O47
• O57
• O53

Acknowledgments

This work was supported by the Tsinghua University research funding of [C2018053] and [20155010298] to Professor Jinghai Zheng.

Correction Statement

Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/mree.This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1. For more discussions, see, respectively, Demurger et al. (2002), Demurger (2001), Zhang and Zhang (2003), Kanbur and Zhang (2005), Wan, Lu, and Chen (2007), Huang, Kuo, and Kao (2003), Yao (1997), DaCosta and Carroll (2001), Jian, Sachs, and Warner (1996), Jiang (2010) and Fleisher, Li, and Zhao (2010).

2. We assume $\phi \left(0\right)=0$. See Hall, Jones, and Hall (1999).

3. A dot over a variable denotes the first-order derivative.

4. Although the dependent variable in (14) pertains to the level, we can always subtract the lagged dependent variable from both sides of the equation to make the left-hand variable one pertaining to the growth.

5. In passing, we should note that it is highly questionable to obtain the value of $\alpha$ by using a regression approach directly based on $lny=\alpha lnk+\left(1-\alpha \right)lnh+\left(1-\alpha \right)A$ (i.e. the production function $y=k^{\alpha }(Ah{)}^{1-\alpha }$) because the endogeneity problem introduced by $lnk$ is difficult to address.

6. We choose the sample period 2004–2016 considering data availability, completeness and consistency.

7. The use of the FE estimator is justifiable in that although the presence of a lagged dependent variable on the right-hand side of equation (14)(14) $y_{it}=\gamma y_{i,t-1}+{\sum }_{j=1}^2{\beta }_j{x}_{it}^j+{\eta }_t+u_i+v_{it}$(14) makes the FE estimator inconsistent when asymptotics are considered in the direction of the cross-section units, the asymptotic properties of panel data estimators can be considered in the direction of time and Arnemiya (1967) has shown that when considered in the direction of time, the FE estimator proves to be consistent and asymptotically equivalent to the Maximum Likelihood Estimator (MLE) (Islam 1995).

8. In passing, it should be noted that the infrastructure variable and the quadradic term are each insignificant when seen separately. However, when taken together, the two terms are jointly very significant. In other words, in Tables 1 and 2 the null hypothesis H₀ that the coefficients on ln(inf_it) and [ln(inf_it)]² are both zero is well rejected, showing that regional infrastructure investment does exert an impact on regional growth.

9. We omit the results of the new regressions here owing to the space limit, as the results contains many statistically insignificant estimates.

10. See Wu (2009, 2011)) for a summary of the capital depreciation rates and initial capital stock levels used in different studies.

11. Our chosen value of $\alpha$, which is $\alpha =0.465$, is quite close to 0.5, which is the value assumed by previous studies such as Chow and Kui-Wai (2002), Chow (2008), Zheng, Angang, and Bigsten (2009), Brandt and Zhu (2010) and Jiang (2011, 2012)).

12. We omit the regression results here to save space, as the results are insignificant.

13. The mechanism is not very clear. This issue deserves further scrutiny. A possible reason may be that poorly developed regions have relatively poor infrastructure and thus are more eager to develop their infrastructure by adding more investment (at least within the sample period). As our analysis in this section is less primary to the central focus of this paper, and given the limited space here, we hope to be able to delve into this issue in our subsequent studies.