The spatial effect of upgrading economic development zones on regional eco-efficiency: evidence from China

Abstract

As the concept of sustainable development gains deeper traction, the upgrading of the economic developmentzone has become a significant way to improve China’s eco-efficiency and resource sustainability. This study uses a spatial econometric model predicated on a three-stage data envelopment analysis model to explore the ramifications of upgrading the economic development zone on regional eco-efficiency. The direct effect evaluation of the economic development zone upgrading shows that it significantly promotes sustainable urban development and urban eco-efficiency. In contrast, the regression results of indirect effects indicate that the economic development zone upgrading policy decrease the eco-efficiency of nearby regions, which is incompatible with sustainable regional development. This spatial effect is realized through the urban level of technological innovation, willingness to regulate the environment, and rising the degree of urban marketization. In addition, in cities with multiple upgraded economic development zones, high government management efficiency, high industrial relatedness, and better transportation infrastructure, the more obvious the direct effect of the policy on the improvement of eco-efficiency in pilot cities and a negative spatial effect on eco-efficiency in surrounding cities.

Appendices

Appendix

The measurement steps of assessment of eco-efficiency

Firstly, the eco-efficiency of all Decision-Making Units (DMUs) was calculated using the conventional DEA model. For any DMU, the input-oriented form of the DEA model can be expressed as Eq. (8):

$$\mathop {\min }\limits_{\theta ,\lambda } \left[ {\theta - \left( {e^{t} s^{ - } + e^{t} s^{ + } } \right)} \right]$$

$$s.t\left\{ {\begin{array}{*{20}l} {\mathop \sum \limits_{i = 1}^{n} \lambda_{i} y_{{{\text{ir}}}} - s^{ + } = y_{ir} } \hfill \\ {\mathop \sum \limits_{i = 1}^{n} \lambda_{i} y_{{{\text{ij}}}} + s^{ - } = \theta x_{{{\text{ij}}}} } \hfill \\ {\mathop \sum \limits_{i = 1}^{n} \lambda_{i} = 1} \hfill \\ {\lambda_{i} \ge 0,s^{ + } \ge 0,s^{ - } \ge 0} \hfill \\ \end{array} } \right.$$

(8)

where i = 1, 2,…, n; j = 1, 2,…,m; r = 1, 2, …, s; n is the number of DMUs, m and s are the input and output variables, respectively. $x_{ij}$ is the jth input element of the ith DMU. $y_{{{\text{ir}}}}$ (r = 1,…,s) is the sth output element of the ith DMU, θ is the effective value of the DMU.

Secondly, by constructing a model similar to the stochastic frontier analysis (SFA), which uses the input slack terms as the dependent variables and the external environment, random disturbance, and management inefficiency terms as the independent variables. The significance of using this model is to increase the input for those decision units that are in a better external environment or have better luck, thus removing the influence of external environmental factors or random factors. We construct Eq. (9) and (10) as follows:

$$s_{{{\text{ik}}}} = \dot{f}\left( {z_{k} ;\beta^{i} } \right) + v_{{{\text{ik}}}} + \mu_{{{\text{ik}}}}$$

(9)

where i = 1, 2,…,n; k = 1, 2,…,m; $s_{{{\text{ik}}}}$ represents the slack variable for the input of the kth DMU i; $z_{k}$ = ($z_{1k}$, $z_{2k}$, …, $z_{{{\text{pk}}}}$) represents p observable external environmental variables; $\beta^{i}$ is the parameter to be estimated of the external environmental variables; $\dot{f}\left( {z_{k} ;\beta^{i} } \right)$ represents the impact of external environmental variables on the input balance value $s_{{{\text{ik}}}}$. $v_{{{\text{ik}}}} + \mu_{{{\text{ik}}}}$ is the mixed error term, $v_{{{\text{ik}}}}$ is random interference; $\mu_{ik}$ means management inefficiency, with an assumption that it obeys a truncated normal distribution, $v_{{{\text{ik}}}}$ and $\mu_{{{\text{ik}}}}$ are not related.

Based on the most effective DMU and based on its input quantity, the adjustment of each other sample input quantity adjusts as follows:

$$\widehat{{x_{{{\text{ik}}}} }} = x_{{{\text{ik}}}} + \left[ {{\text{max}}_{k} \left\{ {z_{k} \beta^{i} - z_{k} \widehat{{\beta^{i} }}} \right\}} \right] + \left[ {{\text{max}}_{k} \left\{ {\widehat{{v_{{{\text{ik}}}} }}} \right\} - \widehat{{v_{{{\text{ik}}}} }}} \right]$$

(10)

where i = 1, 2,…, n; k = 1, 2,…, m; $x_{{{\text{ik}}}}$ represents the actual value of the input of item i of the kth DMU; $\widehat{{x_{{{\text{ik}}}} }}$ is the adjusted value; $\widehat{{\beta^{i} }}$ is the estimated value of the external environmental variable parameters; $\widehat{{v_{{{\text{ik}}}} }}$ is the estimated value of the random interference item.

Finally, by replacing the original input data $x_{{{\text{ik}}}}$ with the adjusted input data $\widehat{{x_{{{\text{ik}}}} }}$ obtained in the second stage, and the output is still the original output data $y_{{{\text{ir}}}}$, we applied the traditional DEA model again to measure the efficiency, which is obtained after excluding the influence of external environmental factors and random factors. The obtained results are used as the dependent variable (Eco-efficiency) of Eqs. (1), (2). The variables used for eco-efficiency calculations are in Table A in the appendix.

Pre-determined variables selection

Industrial structure (IS). Industrial realignment is a crucial development goal for the economy, as it affects the balance between economic growth and environmental protection. Nonetheless, a strong endogenous correlation exists between the industrial structure and the upgrading policy for PEDZ, meaning that the improvement of the regional industrial structure resulting from PEDZ upgrading will, in turn, facilitate the upgrading of PEDZ. To quantify the industrial structure, this study selects IS as one of the pre-determined variables and quantifies the industrial structure using the ratio of tertiary and secondary values added.

Number of PEDZs per city in 2006 (NN). The study aims to determine the impact of PEDZ upgrading on urban ECE. However, the contribution of NEDZ to urban ECE still exists. Therefore, this study identifies the first category of cities mentioned in the selection of independent variables as dummy variables. Specifically, if a city possessed NEDZ before 2006, it is assigned a value of 1; conversely, if it did not, it is assigned a value of 0. Subsequently, this variable is incorporated into the regression model as a pre-determined variable.

Foreign direct investment (FDI). According to the pollution paradise hypothesis, FDI intensifies environmental deterioration by relocating polluting enterprises and industries. The pollution halo hypothesis suggests that FDI can mitigate environmental pollution by introducing environmentally-friendly technologies. It is clear that FDI impacts both economic and environmental development. Thus, this study employs the logarithm of annual FDI in cities as a proxy variable for FDI.

Control variables selection

Retail sales of social consumer goods (SCG). Firms’ production preferences are influenced by consumer demand, which impacts the assessment of ECE. To measure the SCGs, this study employs the natural logarithm of SCGs as the proxy variable.

Urban environmental investment. Infrastructure directly or indirectly affects SDGs. To be specific, infrastructural development can effectively reduce pollution and promote economic development. Therefore, this study uses the logarithm of green space area (UGI) to measure the level of urban environmental development.

Population density (den). Agglomerations of urban populations affect the development of urban economies and pollution, which in turn impact ECE. In addition to creating economies of scale and reducing costs, population concentration improves the quality of the environment . Therefore, this study uses population per square kilometer to measure population density.

The number of provincial-level development zones that have not been upgraded (PN). Although there are many differences in policy content between non-upgraded and upgraded PEDZs, the effects of non-upgraded PEDZs on urban ECE still cannot be ignored. Therefore, this study controls the number of PEDZs (PN) owned by each city per year as the control variable.

Urban education level. The educational resources of a city play an essential role in the city’s economic development. A wealth of educational resources helps cities develop high-level talent and increase their level of innovation. Therefore, this study uses the number of urban universities (uni), urban education investment (edu), and urban research investment (sci) as proxies for the level of urban education.

See Table 8, 9.

Table 8 Global Moran’s index of GTFP

Table 9 The regression results of Eq. (2)