Research on the impact of China's new urbanization on industrial water utilization efficiency – based on spatial spillover effects and threshold characteristics

The research in this article focuses on the following two aspects: first, the existing studies on the relationship between NUD and IWE have not established a complete theoretical framework, resulting in a lack of theoretical support for related re-examination. Therefore, this article constructed a theoretical framework between NUD and IWE under the premise of considering spatial spillover effects and threshold effects. Second, the existing literature mainly has adopted linear analysis to explore the impact of social and economic factors on IWE, but lacked a nonlinear perspective to analyze the impact of NUD on IWE. In addition, due to the differences in the levels of NUD and IWE in different regions of China, the hypothesis of spatial homogeneity in previous studies has been interpreted differently for the effects of NUD. Therefore, this article attempted to analyze the impact of NUD on IWE from the perspective of nonlinear and spatial spillover characteristics on the basis of existing research findings, so as to provide experiences and insights for promoting regional cooperation and achieving sustainable development.

The rest of this article can be divided into five sections. Section 2 presents the literature review. Section 3 presents the theoretical framework. Section 4 introduces the models along with the selection of variables and data sources. Section 5 presents the empirical analysis of the threshold effect and spatial spillover effect of NUD on IWE. Section 6 presents the conclusions and policy recommendations.

Urbanization is defined as the share of the urban population in the total population.

In recent years, with the growing prominent conflict between water supply and demand, water resources have become the major bottleneck limiting sustainable economic development. As the second largest user of water resources in China, industrial water consumption is under the pressure of water shortage and will increase during the ‘14th Five-Year Plan’ period as the number and scale of industrial enterprises continue to increase. Therefore, the issue of IWE is not only of great concern to the government but also of increasing interest to the academic community. The existing studies on IWE in the literature focus on two aspects: first, the measurement of IWE; and second, the influencing factors of IWE.

Data envelopment analysis (DEA) and the stochastic frontier analysis (SFA) have been chosen by most researchers to measure IWE. The DEA model was proposed by Charnes and has been widely used to calculate the efficiency of resource utilization. Cabrera et al. (2018) calculated the efficiency of 194 water utilities in Portugal by using the DEA model and provided recommendations for inefficient DMUs (decision making units). Chen et al. (2021) measured IWE in 31 Chinese provinces from 2005 to 2015 by using the self-sampling DEA method. Although the DEA model can measure the efficiency of ‘multi-input’ and ‘multi-output’, it ignores the effect of random errors. However, the SFA model takes into account the effect of the presence of random errors on the results, making the calculation more accurate. Lei et al. (2017) measured the IWE by using the SFA model on the basis of 31 provinces of China from 2004 to 2014. Zheng et al. (2018) introduced the shepherd water distance function to construct the total factor water efficiency index and then calculated the efficiency index by using the SFA model. The SFA model considers the effect of the presence of random errors on the results, but it may cause structural bias due to the misconfiguration of the production function. Therefore, some more advanced models such as the slack-based model (SBM) were proposed based on traditional models to solve the problem of inaccurate calculation results due to the shortcomings of traditional efficiency methods. The SBM model addressed the issue of how to deal with undesirable outputs, which allows scholars to conduct more in-depth research. Ding et al. (2019) measured the IWE of the Yangtze River Economic Belt by using the improved super-efficient DEA model. Liu et al. (2020) measured IWE in China by using the SBM model based on undesirable outputs. However, the SBM model lacks information on the ratio of input or output target values to actual values, so the epsilon-based measure (EBM) model proposed by Tone and Tsutsui (Tone & Tsutsui 2010) addressed this problem. The EBM model includes both radial and nonradial slack variables, which makes the calculated efficiency values closer to the actual situation and has been adopted by most scholars in efficiency calculations (Wang et al. 2017; Liu & Peng 2020).

It can be found that existing studies have mostly used radial distance functions such as CCR (A. Charnes, W.W. Cooper and E. Rhodes) and BCC (Banker, Charnes, and Cooper) or nonradial distance functions such as SBM, but they all have shortcomings. The EBM model is a hybrid model that integrates both radial and nonradial distance functions, and it not only accounts for slack variables but also solves the problem of losing the original scale information of the projected value of the efficiency frontier, thus making the calculation results more accurate.

There are many relevant studies on the relationship between urbanization and IWE, but the results are quite different. From the perspective of linear analysis, most scholars believe that urbanization can improve IWE. Zhao et al. (2022) concluded that the increasing level of urbanization has a promotion effect on improving the water environment by using 43 prefectural units in the Yellow River Basin of China as the research area. Li et al. (2019) found a positive effect of urbanization on IWE using data from 31 Chinese provinces as a sample. Zheng et al. (2018) and Bao et al. (2016) investigated the relationship between urbanization and IWE and came to similar conclusions. Some scholars also argue that urbanization inhibits IWE. Ding et al. (2019) found that both population and land urbanization have a negative impact on IWE using a sample of provinces in China's Yangtze River Economic Belt. Hai et al. (2018) found that urbanization has a negative impact on IWE by analyzing the relationship between urbanization quality and IWE from different urbanization perspectives, such as population, economy, society, and land. Other scholars believe that urbanization has different impacts on IWE in different regions due to regional differences (Ma et al. 2018; Liu et al. 2019).

From the perspective of nonlinear analysis, as many economic variables in reality have structural mutation problems, especially for large samples and panel data, thus the PTM has been widely adopted by scholars to find structural mutation points for nonlinear studies. The PTM was proposed by Hansen (1999), in which the static PTM has been used to study the relationship between variables. Guo et al. (2022), Huo et al. (2022), Li et al. (2020), Luo & Zhang (2021), and Wu et al. (2021) investigated the threshold effects between green economy or innovation efficiency and its influencing factors by using the static PTM. Yang et al. (2019) and Zhang et al. (2019) studied the threshold effects between eco-efficiency and its influencing factors. It can be seen that there is a lack of nonlinear studies on the impact of urbanization on IWE, but some scholars have studied the nonlinear relationship between urbanization and water resources. Wang et al. (2022) studied the threshold effects of urbanization and industrial structure change on water stress by using the PTM. Zhang et al. (2011) found a complex quadratic curve relationship between urbanization and water use in Urumqi.

From the perspective of spatial spillovers, the spatial panel data models were used by scholars to study spatial characteristics, with application areas ranging from regional economics to real estate economics and financial geography, etc. (Zhang 2016; Wang & Guan 2017; Iwata et al. 2019). Not only that but also the traditional fields of labor economics (Azorin & de la Vega 2015), energy economics (Zhang et al. 2017; Pan et al. 2018; Kiziltan 2021), environmental economics (Huang et al. 2020; Karahasan & Pinar 2022), international trade (Metulini et al. 2018), and other aspects of research have risen to the spatial level. Although the spatial spillover effects of urbanization on IWE are relatively scarce, Zhang et al. (2020) found that urbanization has spatial spillover effects on water utilization efficiency and a negative effect on agricultural water utilization efficiency. Lu et al. (2022) constructed the dynamic spatial econometric model, and the spatial decomposition results showed that urbanization has a short-term positive effect on agricultural water utilization efficiency in the region and surrounding areas. We found no consensus among scholars regarding the spatial impact of urbanization on water utilization efficiency.

The choice of empirical research method has a great influence on the research results. Existing studies only have analyzed the impact of urbanization on IWE from a single aspect of spatial spillover effects or threshold effects, which may lead to bias in the estimation results. Therefore, this article integrated the research findings of spatial spillover effects and threshold effects and systematically analyzed the impact of NUD on IWE.

New urbanization is a comprehensive development process, which involves economic development, ecological environmental protection, social governance, and other aspects. Moreover, new urbanization, as a complex economic variable, is likely to have structural abrupt points in its development process, and thus, its impact on IWE may be nonlinear. Therefore, this article studied the threshold effects of NUD on IWE from the following aspects.

The NUD can affect IWE through the pooling of production factors such as population and resources. (1) The NUD causes a large number of people to move from rural to urban areas, increasing the population density and expanding the size of cities. Moreover, the population tends to move to areas with higher levels of economic development and more opportunities in the process of population transfer (Feng et al. 2022). The influx of high-tech talents continuously improves the level of science and technology, thereby enhancing regional innovation capabilities, reducing industrial water consumption, and improving industrial water resource reuse and IWE. However, excessive population gathering may also lead to urban development problems, such as a sharp increase in domestic water consumption and pollution of the water environment, thus resulting in inadequate and inefficient industrial water utilization. (2) The NUD optimizes resource allocation and improves industrial productivity and IWE by providing favorable conditions for the pooling of various production resources (Fujita & Mori 1997).

The NUD can affect IWE by increasing the demands of residents on the living environment and strengthening the management and control of industrial enterprises. (1) The industrial enterprises have to transform and upgrade their production methods and purchase more advanced wastewater purification equipment to meet the government requirements for pollution control and reduction of wastewater discharge, which lead to an increase in production costs (Bo 2021). Meanwhile, industrial enterprises carry out innovation of production technology to save costs, that is, to make the benefits by implementing technological innovation offset the increased costs of enterprises, thereby improving IWE. (2) In terms of lifestyle, the NUD makes residents more aware of environmental protection and therefore a higher demand for urban environmental quality, thus forcing enterprises to reduce the discharge of wastewater and sewage and improving IWE. In addition, a series of environmental protection measures by industrial enterprises improve the urban environment and attract more talents to settle down and therefore help increase labor productivity and IWE. However, the improvement of living standards and changes in lifestyles caused by NUD make residents increase their demand for service industries, while leading to the rapid development of the tertiary industries, thus resulting in affecting the optimization and upgrading of industrial structure and hindering the improvement of IWE (Fox 2012).

The NUD can affect IWE by optimizing the water use structure of industrial enterprises and adjusting the industrial structure. (1) The new urbanization advocates environmental protection and resource conservation. Therefore, industrial enterprises improve the reuse rate of water through cooling water recycling, sewage reuse, and other methods to reduce the consumption of fresh water and then optimize the structure of water use. (2) The new urbanization requires the efficient and economical utilization of resources, which promotes industrial enterprises to vigorously develop water-saving intensive industries and abandon water-intensive industries that use water in a sloppy manner, thus making the industrial structure optimized and helping to improve IWE.

The NUD can affect IWE by increasing the number of industrial enterprises. The degree of industrialization is becoming higher and higher to adapt to the rapid development of new urbanization, which makes the number of industrial enterprises continue to increase and expand the scale of production, thereby increasing the consumption of industrial water and inhibiting the improvement of IWE (Balland et al. 2020). But the expansion of regional production scale may also reduce production costs and help to improve IWE.

Accordingly, this article proposed research hypothesis 1: the NUD has threshold effects on IWE.

The NUD creates a platform for the inflow of various resources. The rapid development of new urbanization in the region attracts a large inflow of capital, advanced technology, and talents, which lays the foundation for innovation and therefore improves the IWE in the region (Lahr 2009). Meanwhile, the labor force in the surrounding areas flows into the city on a large scale due to the profit-seeking psychology, and various production factors from the surrounding areas also keep flowing in, thus weakening the ability of the surrounding areas to improve IWE. On the other hand, technological innovation in this region can generate knowledge spillovers and spread to the surrounding areas, so the surrounding areas can make technological changes through imitation and secondary innovation to improve IWE.

The NUD promotes the development of water-saving intensive industries, and high-efficiency enterprises cluster in cities due to the selection effects, so water-intensive and low-efficiency enterprises with sloppy water consumption are pushed out to the surrounding areas. Then the high-efficiency enterprises clustered in cities can continuously reduce production costs due to the scale effects, exacerbating the gap with surrounding areas (Yuan et al. 2022), that is to say, it is not conducive to the improvement of IWE in the surrounding areas. On the other hand, the NUD focuses on regional coordinated and balanced development. Neighboring cities can generate complementary industrial clusters through the division of labor and cooperative production and share advanced production technologies, which can improve IWE in the surrounding areas.

Since the purpose is ‘people oriented’, the new urbanization focuses on the construction of infrastructure such as culture, education, medical care, and transportation, which improves the convenience and welfare of residents and attracts the inflow of capital and talents. Then the influx of innovative talents can improve the productivity of enterprises and expand the advantages of the region, without taking advantage of the improvement of IWE in the surrounding areas. On the other side, neighboring areas can form close-knit city clusters due to the improvement of transportation, communication, and other infrastructures, which is conducive to accelerating knowledge dissemination and information sharing (Qin 2017) and the development of the surrounding areas. In addition, the NUD of the region promotes the environmental awareness of residents in the surrounding areas by influencing their lifestyles (Liddle 2004), thus forcing industrial enterprises to upgrade their technology and helping to improve IWE.

Accordingly, this article proposed hypothesis 2: NUD has spatial spillover effects on IWE.

IWE_it is the industrial water efficiency of the ith region in the t year; NUD_it is the NUD of the ith region in the t year; PGDP, PCWR, TWR, TP, ISF, ISS, ER, and FDI are the control variables of this articles; the specific description is shown in Table 3; γ₁–γ_n+1 are the threshold value; β₁₁–β_1(n+1) are the regression estimation coefficients of different threshold intervals; and I (*) is the indicative function.

The EBM model was used to measure IWE. On the basis of the existing literature (Zheng et al. 2020), the industrial fixed assets, industrial employment, and industrial water consumption of 30 provinces (cities and autonomous regions) in China were selected as input indicators, and industrial added value (expressed at constant prices in 2011) and industrial wastewater discharge are indicators of desirable and undesirable outputs, respectively. The indicator system is shown in Table 1.

The core explanatory variable is NUD, which was measured by the entropy method. Referring to the existing literature (Wu et al. 2019; Ma et al. 2021), a comprehensive evaluation system of NUD composite indicators was established from the five aspects of population, economy, ecology, society, and urban and rural development, which was composed of 14 specific indicators, as shown in Table 2.

(1) Economic development level, measured by GDP per capita. (2) Natural resource factors, measured by water resources per capita (water endowment) and total water resources. (3) Technological progress, measured by the number of domestic patent applications granted. (4) Industrial structure, measured by the tertiary industry as a share of the output value of the secondary industry and the share of tertiary industry output in GDP. (5) Environmental regulation, measured by completed an investment in industrial wastewater treatment. (6) Foreign direct investment. Since GDP per capita, completed investment in industrial wastewater treatment and foreign direct investment have an exponential growth trend, they are treated as logarithms.

In this study, 30 provincial-level cities in China were selected as research objects. Due to the lack of data in Hong Kong, Macau, Taiwan, and Tibet, this study does not include the aforementioned four cities for the time being. The research period range is 2011–2020. The data in this article come from China Statistical Yearbook, China Environmental Statistical Yearbook, and China Industrial Statistical Yearbook. Explanations of specific indicators and statistical analysis of the data are shown in Table 3.

Table 4 shows the correlation matrix of the variables. It can be seen that most of the variables are correlated at the 1% significance level. Moreover, the correlation coefficient between the explained variable IWE and the core explanatory variable NUD is 0.328, which is significantly positive at the 1% level, indicating that a high level of NUD is conducive to improving IWE.

The regression results of the PTM are shown in Table 7. The linear estimation results of the ordinary least squares (OLS) are also given for the purpose of comparative analysis. As shown in column (1) of Table 7, the estimated coefficient of NUD by OLS is significantly positive, indicating that the NUD is beneficial for improving IWE. According to column (2) of Table 7, we found that the NUD has a significant single threshold effect on IWE, that is, when the level of NUD is lower than 0.3137, it has a positive but not significant effect on IWE, and when the level of NUD is higher than 0.3137, its regression estimated coefficient is 0.540 and is significantly positive at the 5% level. The possible explanation is that there is a bottleneck in the NUD, and the improvement effect of the lower level of NUD on IWE is not significant, while with the improvement of the level of NUD, the industry actively carries out technological transformation and industrial structure upgrading and pays more attention to resource protection and coordinated development of industry and ecological environment in the production process, so that the negative externalities caused by ‘urban disease’ are gradually alleviated, which is conducive to the improvement of IWE. In conclusion, the linear model does not fully reveal the effect of NUD on IWE, so hypothesis 1 was confirmed.

From the perspective of the control variables, the rise of PGDP is conducive to the improvement of IWE, while ISF has a negative inhibitory impact on IWE. From the regression coefficients, the coefficient of PGDP is the largest, which indicates that it has the most obvious impact on IWE, so improving the level of economic development can significantly promote the improvement of IWE.

The global Moran's I was used to examine the spatial autocorrelation of IWE and NUD. As shown in Table 8, the global Moran's I values of NUD from 2011 to 2020 show an overall upward trend from 0.241 in 2011 to 0.305 in 2020, and all are significant at the 1% level, indicating that it has a significant positive spatial autocorrelation. The global Moran's I values of IWE are positive from 2011 to 2020, and most of them are significant at the 10% level. Overall, the IWE has a positive spatial autocorrelation.

The Hausman test was first performed and it passed the 1% significance test, so the fixed effect model was selected. Next, the Lagrangian multiplier test (LM test) was carried out, and the LM error, robust LM error, and robust LM lag all passed the 1% significance test and the LM lag passed the 10% significance test, so the SDM of fixed effect was chosen. Finally, the LR test and Wald test were carried out, and both passed the 1% significance test, indicating that the SDM cannot degenerate into SAR or SEM. The results are shown in Table 9.

The estimated results of the SDM are shown in Table 10. We found that the time fixed effect is most appropriate by comparing. According to the results, the autoregressive coefficient is 0.184 and is significant at the 5% level, which means that there is a significant positive spatial spillover effect on IWE. The regression estimation coefficient of the core explanatory variable NUD is 0.666 and is significant at the 1% level, indicating that NUD is conducive to the improvement of IWE. The possible explanation is that the construction of new urbanization has brought a large amount of resources to the cities, which are expanding in size while promoting economic development and technological progress, thus enabling industries to accelerate the transformation and upgrading of their industrial structure and technological innovation. Furthermore, the policy changes brought about by the NUD have forced industrial enterprises to abandon water-intensive industries and vigorously develop water-saving industries. In addition, the construction of new urbanization has improved the quality of the population, enhanced awareness of water conservation and environmental protection among residents, and improved the construction of social infrastructure in terms of education, social functions, and environmental protection, thereby making industrial enterprises pay more attention to issues such as water recycling and wastewater discharge.

Since the regression estimation coefficients in the SDM cannot reflect the impact of NUD in the adjacent areas on the IWE of the region, this article decomposed the spatial effects. As shown in Table 11, we report the decomposition results of both the spatial adjacency matrix and the geographic distance matrix. We can see that the estimated coefficient of the direct effect of NUD is significantly positive at the 1% level, confirming that the NUD can indeed improve IWE by optimizing the industrial structure, rationally allocating resources, and strengthening technological innovation. The estimated coefficient of the indirect effect of NUD is 1.119, which is significant at the 5% level, confirming that there is a significant positive spatial spillover effect, that is, NUD can not only directly improve IWE in this region but also provide support for the improvement of IWE in this region by influencing NUD in the surrounding areas. In addition, from the perspective of the estimated coefficients, the estimated coefficient of the indirect effect of NUD is higher than the direct effect, which means that the positive driving effect of NUD on the improvement of IWE depends to a large extent on the spatial spillover effects. Hypothesis 2 was, therefore, verified.

From the perspective of control variables, we found that the direct effect of PGDP is significantly positive, and the indirect effect is significantly negative, which means that the increase of PGDP in this region is conducive to the improvement of IWE, but the increase of PGDP in surrounding areas has a negative impact on the IWE in this region. The indirect effect of ER is significantly negative, which is explained by the fact that the enhanced environmental regulations in the surrounding areas have led industrial enterprises to relocate high water-consuming and high-polluting industries to the region, thereby adversely affecting the improvement of IWE in the region. The direct effect of PCWR is significantly negative, that is to say, the abundance of natural resources has hampered the improvement of IWE, which is a typical phenomenon of the ‘resource curse.’ The direct effect of FDI is significantly positive, which is due to the fact that FDI has brought capital, technology, and advanced management concepts to industrial development and promotes the improvement of IWE. Since the direct and indirect effects of ISS are not significant, this article used ISF to represent the industrial structure for analysis. We can see that the direct effects of both ISF and TP are significantly negative, which is consistent with the research of Li et al. (2019), where there is a time lag in the implementation of scientific and technological innovation and the shortcomings of industrial structure, both of which reduce IWE.

This study explored the mechanism of action between NUD and IWE. First, the EBM model and the entropy method were used to measure IWE and NUD. Second, the spatial spillover effects and threshold characteristics of NUD on IWE were discussed by using the SDM and the PTM. The following conclusions are drawn:

• (1)

  The IWE and NUD are unbalanced in regional development and both show a spatial pattern that gradually decreases from east to west. The IWE in the eastern region is significantly higher than that in the central and western regions, and it shows a trend of spreading from the east to the central and western regions. The level of NUD is gradually improving, and cities with a high level of development are mainly distributed in the eastern region, while there is a large room for improvement in the central and western regions.

• (2)

  The NUD has a significantly positive spatial spillover effect on IWE. Both the direct and the indirect effects of NUD are significant positive by using the partial differential method to decompose the total effect, that is, the NUD can not only improve IWE in this region but also promote IWE by influencing NUD in the surrounding areas. Furthermore, the spillover effects between regions are greater than the local effect within the region.

• (3)

  There is an interval effect between NUD and IWE, and the structural change point is 0.3137. The results of the panel threshold regression model show that when the level of NUD is below the threshold value, the promotion effect of NUD on IWE is not significant; when NUD exceeds the threshold value, the driving effect is significantly increased.

On the basis of aforementioned conclusions, this article put forward the following policy recommendations:

First, the government should pay attention to regional differences to formulate policies and plans that are in line with local development according to local conditions. Beijing, Zhejiang, Guangdong, and other high-efficiency cities can maintain their current resource management policies and investments. Medium-efficiency and low-efficiency cities should actively develop their economies and strengthen water resources management. The government should try to reduce the number of high water-consuming and high-polluting enterprises in the industrial layout. Moreover, areas with slow development of new urbanization should pay attention to infrastructure construction and improve the quality of public services to make up for the shortcomings of development.

Second, we should strengthen the construction of new urbanization, vigorously improve the quality of urbanization development, and promote industrial development. The regions should cultivate a sense of cooperation, strengthen exchanges and cooperation between regions, and bring into play the demonstration and radiation role of urban agglomerations. Moreover, the regions should strengthen the deep integration of new urbanization and technological innovation to create a good environment for the full release of technological spillovers brought about by the development of new urbanization, so as to achieve an increase in the efficiency of industrial water utilization.

This research was supported by National Natural Science Foundation of China (Grant No.41701613) and National Natural Science Foundation of China (Grant No.42271303).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.