Introduction

The Pearl River Delta Urban Agglomeration (PRDUA) has experienced tremendous growth during China’s 40 years of reform and opening-up, achieving remarkable urbanization milestones. In 1978, the urbanization rate of the region’s resident population was only 16.26%1. By 2021, it had surged to 84.85%, representing an average annual increase of 1.04%2. This rapid urbanization has been accompanied by significant urban land expansion, with the built-up areas in municipal districts increasing from 2012 km2 in 2006 to 4113 km2 in 2021. The average annual growth rate of urban construction land between 2006 and 2021 exceeded 6.72%, nearly 1.5 times the national average3,4. However, such rapid expansion has led to inefficiencies in land use, with substantial areas of urban land remaining underutilized5. Given PRDUA’s status as one of China’s three major urban agglomerations, there is an urgent need to explore its land use efficiency to inform development strategies and optimize land allocation policies.

This paper applied calculate the urban land utilization eco-efficiency (ULUEE) of nine cities in the Pearl River Delta region, subsequently analyzing the influencing factors of ULUEEThe contributions of this paper are threefold: (1) It introduces the concept of urban land use eco-efficiency (ULUEE) and provides a detailed explanation of its definition. (2) It applies the super-efficiency EBM model with undesirable outputs to measure the ULUEE of nine cities in the PRDUA, which has three advantages: firstly, combining both radial and non-radial factors; secondly, considering undesirable outputs; thirdly, making it possible that the efficiency value is more than 16. (3) A Tobit regression analysis is conducted to identify the factors influencing ULUEE, providing policy recommendations for government decision-making. The Tobit model is known as the truncated regression model, which can handle metrics demonstrating both partially continuous and partially discrete distributions of the dependent variable7. The research framework of this issue is shown in Fig. 1.

Fig. 1
figure 1

Flowchart of the empirical research of ULUEE in China.

Full size image

Literature review

Recent research has focused on evaluating urban land use efficiency, primarily using data envelopment analysis (DEA)8,9,10,11,12,13,14,15,16,17,18,19,20,21 and stochastic frontier analysis (SFA)22,23,24,25,26,27,28. However, these approaches present certain limitations. First, the selection of evaluation metrics often neglects ecological and environmental factors. Second, the SFA method, which relies on parameter estimation, assumes independence among the variables—a condition that is rarely met in practice29,30. Traditional DEA models, such as CCR, BCC, and SBM, fail to simultaneously account for both radial and non-radial characteristics, potentially biasing the results31,32,33,34,35,36,37,38.

To address these gaps, this paper incorporates energy consumption and pollutant emission indices into the urban land use efficiency evaluation framework. By integrating socioeconomic and environmental benefits, this approach allows for a more comprehensive and accurate assessment. Moreover, this study pioneers the application of a super-efficiency epsilon-based measure (EBM) model with undesirable outputs to assess urban land use eco-efficiency (ULUEE) in the PRDUA. The model has three key advantages: it considers both radial and non-radial characteristics in technical efficiency evaluation, accounts for undesirable outputs, and allows for efficiency scores exceeding 16,39,40, thereby enhancing the precision of the evaluation.

Data and methodology

Definition of urban land use eco-efficiency

Schaltegger and Stum first put forth the concept of eco-efficiency in 1990, defining it as the ratio of economic growth to environmental impact41. The word “eco-efficiency” contains the root of the word “economy” as well as that of the word “ecology”. Therefore, “eco-”envelops the meanings of both economy and ecology. Despite different understandings of eco-efficiency, most scholars accept that it entails achieving maximal economic gains with minimal resource consumption and environmental costs42,43. Based on comprehensive and scientific principles, this paper defines urban land use eco-efficiency as follows: under the condition of stable or decreased input of productive factors, the overall production system in the built-up area in a city municipal district achieves greater economic gains at less cost to the environment.

Methodology

Super-efficiency EBM model with undesirable outputs

This paper applies a super-efficiency EBM model with undesirable outputs to calculate ULUEE in the PRDUA. There are n DMUs and each DMUj (j = 1,2 …,n) applies m inputs to generate s desirable outputs and q undesirable outputs. xj = (x1j, x2j, …, xmj)T, yj = (y1j, y2j, …, ysj)T and kj = (k1j, k2j,…, kqj)T stand for the column vector of inputs, desirable outputs and undesirable outputs; The model can be represented as follows44:

$$\begin{aligned} & \gamma^{*} = \min \left( {\frac{{\beta - \varepsilon x\sum\nolimits_{i = 1}^{m} {\frac{{\omega_{i}^{ - } s_{i}^{ - } }}{xio}} }}{{\delta + \varepsilon y\sum\nolimits_{r = 1}^{s} {\frac{{\omega_{r}^{ + } s_{r}^{ + } }}{yro} + \varepsilon_{k} \sum\nolimits_{p = 1}^{q} {\frac{{\omega_{p}^{ - } s_{p}^{ - } }}{{k_{po} }}} } }}} \right) \\ & s.t\left\{ {\begin{array}{*{20}c} {\sum\nolimits_{j = 1j \ne 0}^{n} {x_{ij} } \lambda_{j} + s_{i}^{ - } = \beta x_{io} } & {i = 1,2,...,m} \\ {\sum\nolimits_{j = 1j \ne 0}^{n} {y_{rj} } \lambda_{j} - s_{r}^{ + } = \delta y_{ro} } & {r = 1,2,...,s} \\ {\sum\nolimits_{j = 1j \ne 0}^{n} {k_{pj} } \lambda_{j} + s_{p}^{ - } = \delta k_{po} } & {p = 1,2,...,q} \\ {\lambda_{j} \ge 0,s_{i}^{ - } \ge 0,s_{r}^{ + } \ge 0,s_{p}^{ - } \ge 0} & \; \\ \end{array} } \right. \\ \end{aligned}$$
(1)

where \(\eta^{ * }\) denotes the value of ULUEE; β and δ are the parameters of radial measure;. n, m, s, and q represent the number of DMUs, the inputs, the outputs, and the undesirable outputs, respectively. \(xio\), \(yro\) and \(kpo\) stand for the ith input, rth desirable output and pth undesirable output of the oth DMU (afterward DMUo), respectively; \(s_{i}^{ - }\), \(s_{r}^{ + }\) and \(s_{p}^{ - }\) indicate the slack variables of the ith input, rth desirable output and pth undesirable output, respectively. \(\omega_{i}^{ - }\),\(\omega_{r}^{ + }\), and \(\omega_{p}^{ - }\) stand for the weights of the ith input, rth desirable output and pth undesirable output, respectively;. \(\varepsilon_{x}\), \(\varepsilon y\) and \(\varepsilon k\) are the ckey parameters tha signify the non-radial weights of inputs, desirable outputs and undesirable outputs, respectively; λi represents the linear combination coefficient of DMUj.

Tobit model

In this paper, the ULUEE values is larger than 0, it is a truncated regression problem. If the ordinary least squares method is applied to make empirical analysis on the relationship of the influencing factors with ULUEE, the parameter estimation results may get the biased45,46,47. The Tobit model can effectively avoid this problem with the Maximum Likelihood Estimation48,49,50,51. Therefore, the Tobit regression model is used in this study. The form of the Tobit model is as follows:

$$Y = \left\{ {\begin{array}{*{20}c} {Y^{ * } = a + X\beta + \varepsilon \begin{array}{*{20}c} {} & {Y^{ * } \ge 0} \\ \end{array} } \\ {\begin{array}{*{20}c} 0 & {\begin{array}{*{20}c} {} & {} \\ \end{array} } & {\begin{array}{*{20}c} {\begin{array}{*{20}c} {\begin{array}{*{20}c} {} & {} \\ \end{array} } & {} \\ \end{array} } & {Y^{ * } { < 0}} \\ \end{array} } \\ \end{array} } \\ \end{array} } \right.$$
(2)

where Y represents the interpreted variable. X stands for the explanatory variable; α denotes the intercept vector; β is the parameter vector; ε indicates the normal distributed error term, ε ~ N(0,σ2).

Data source and indicator selection

This paper’s research area is the PRDUA, which contains Guangzhou, Shenzhen, Foshan, Zhuhai, Huizhou, Dongguan, Zhongshan, Jiangmen, and Zhaoqing (Fig. 2). By the end of 2021, the total population of the Pearl River Delta region was 57.2 million; regional GDP reached 731,877 billion yuan, accounting for 9.83% of national GDP. The study time period is from 2006 to 2021.

Fig. 2
figure 2

The PRDUA of China.

Full size image

Consulting the results of previous research9,11,12,14,16,52,53, we chose the urban construction land area, urban capital stock, and number of employees in secondary and tertiary industries as the input indicators; secondary and tertiary industries added values as desired outputs; and urban industrial SO2, and urban industrial soot as undesired outputs (Table 1). This paper applies the perpetual inventory method54,55 to estimate the urban capital stock in the PRDUA. The calculation method is: Ki,t = Ii,t + (1 − δ) Ki,t−1. K is urban capital stock. I stands for urban fixed asset investment. δ denotes the depreciation rate of urban capital stock. The subscripts i and t represent city and year, respectively. Referring to Zhang et al.56, we set δ equal to 9.6%. The urban capital stock in 2006 is equal to the urban fixed asset investment in 2006 divided by 10%. All data were collected from the Guangdong Statistical Yearbook (2007–2022)4.

Table 1 ULUEE measurement index system.
Full size table

Analyzing the characteristic of ULUEE

According to Eqs. (1), we calculated the ULUEEs of nine cities in the PRDUA from 2006 to 2021, which are shown in Table 2, Fig. 3, and Fig. 4.

Table 2 The ULUEEs of nine cities in the PRDUA from 2006 to 2021.
Full size table
Fig. 3
figure 3

The average ULUEE values of nine cities in the PRDUA from 2006 to 2021.

Full size image
Fig. 4
figure 4

Evolutionary trend of the Mean ULUEE and variation coefficient of ULUEE in nine cities in the PRDUA from 2006 to 2021.

Full size image

Spatial characteristics of ULUEE

As shown in Table 4, the annual average ULUEE value of Shenzhen, Guangzhou, and Foshan appear to be much higher, at 1.028, 1.038, and 1.091 respectively. There are several possible explanations for understanding such results. As the growth pole of the region, Shenzhen and Guangzhou have boasted nearly perfect infrastructure construction over the past years. With strong industrial foundation and rich human capital, as well as the extremely high concentration of colleges and universities, these cities have occupied significant advantages for independent innovation. There is no doubt that more advanced technologies and perfect hardware conditions will make great contributions to improving their level of ULUEE. In addition, these advantages not only benefit the local development, but also usually have regional spillover effect on cities adjacent to the growth pole. As a result, Foshan is probable to receive the positive effects of industrial transitions from Guangzhou and Shenzhen, such as improving the internal infrastructure and optimizing the production technology, which are conclusive to maintain a high-level ULUEE value. Meanwhile, the ULUEE level of Zhuhai, Dongguan, Zhongshan, and Jiangmen are relatively lower, with all the annual average ULUEE values being less than 0.6 and the lowest value in Dongguan only 0.453, which indicates a large amount of excessive input of land resources for economy development.

Temporal characteristics of ULUEE

As shown in Table 2 and Fig. 3, the average value of ULUEE in the Pearl River Delta region from 2006 to 2021 as calculated with the super-efficiency EBM model with undesirable outputs ranges between 0.724 and 0.782, and the annual average value is 0.745. Among them, ULUEE was low between 2011 and 2014, with the highest point was in 2007 and the lowest point was in 2014. As for the variation coefficient of ULUEE, an undulate variation was shown. It is worth noting that the variation coefficient showed an overall downward trend after 2017, impling that the spatial gap of ULEE between urban agglomerations tended to decrease after 2017. With the concept of green development gradually gaining popularity and the continuous promotion of ecological civilization construction, the ULUEE in Shenzhen, Zhuhai, Foshan, Dongguan and Jiangmen compared with the initial research stage has been significantly improved. In contrast, the ULUEE in Zhongshan city has decreased considerably, which may be related to the problems existing in resource utilization, environmental protection or urban development strategy in the city, which needs to an in-depth analysis according to specific situations for effective measures.

Analyzing the influencing factors on ULUEE

First, it is certain that a change in the potential value of ULUEE is caused by the combined action of multiple factors. In order to investigate the reasons, we conducted a Tobit regression and analyzed systematically the effect of government intervention in the economy (GIE)57,58,59, economic development level (EDL)60,61,62, industrial structure (IS)63,64,65,66, opening-up level (OUL)67,68, green technological progress level (GTPL)64,69,70, and infrastructure construction level (ICL)71,72,73 on ULUEE (Table 3).

Table 3 Influencing factors of ULUEE.
Full size table

Determinants of ULUEE

Government intervention in the economy

The economic development of China cannot be separated from government intervention. Government expenditure directly stimulates economic growth through government consumption and government investment57, and it directly affects the economic output of land. The scale of government expenditure reflects the ability of government to macro-economic regulation. On the one hand, fiscal expenditure can better support the construction and maintenance of all kinds of infrastructures, and improve the infrastructure environment, which plays a very important role in improving land utilization efficiency58. On the other hand, the optimal allocation of land resources cannot implement the realization of the optimal allocation of land resources cannot exist without the market mechanism59. The excessive intervention of the government may restrict the normal market mechanism, which would reduce the efficiency of land resource allocation. Therefore, the impact of government intervention on ULUEE must be tested.

Economic development level

The higher the level of urban economic development, the stronger the agglomeration effect of capital, labor, technology and other factors on urban area, which in turn promotes a gradual intensification of urban land utilization. Xie et al.60, Yu et al.61, and Ma et al.62, all believe that higher economic development levels produce high efficient land resource allocation. Therefore, this paper expects to find a positive correlation between economic development level and ULUEE.

Industrial structure

The optimization of industrial structure is mainly manifested in on industrial intension, intensification of knowledge economy service, increased degree of industrial correlation, extension of the industrial chain. Referring to the research of Zhu et al.63, Yu et al.64, and Fan et al.65, this paper uses the ratio of tertiary industry output to second industry output to represent industrial structure and expects the industrial structure to have a positive impact on ULUEE.

Opening-up level

By raising the level of opening up, domestic enterprises can learn about advanced international technology production technology and management experience, and promote the transformation from extensive to intensive development model of urban construction land utilization. Referring to previous research results67,68, this paper applies the ratio of total export–import volume to GDP to represent opening up level, and expects urban levels of opening up to have a positive impact on ULUEE.

Green technological progress level

Green technological progress can inhibit the pollutant emissions and mitigate the contradictions between man and ecological environment caused by the rapid expansion of urban land. The number of green patent authorizations in a year can reflect local technological progress level. Making reference to the research results of Yu et al.64, Jin et al.69 and Zhao et al.70, this paper expects technological progress to have a positive impact on ULUEE.

Infrastructure construction level

In areas with better infrastructure such as transportation, the returns of urban land is often higher, but the repeated construction and inefficient operation of infrastructure leads to waste of land and various resources, deterioration of the ecological environment. Therefore, with reference to Han and Lai71, Luo and Peng72, Wang et al.73, and Li et al.66 this paper argues that the impact of infrastructure construction level on ULUEE requires testing.

Explaining ULUEE: tobit regression results

In this section, we make empirical analysis on the influencing factors of ULUEE. Firstly, we conduct the tests of Correlation and Variance Inflation Factor (VIF) to avoid the existence of the problem of multicollinearity among variables66,74,75,76. Table 4 shows the correlation coefficient matrix of all variables. The maximum value of correlation coefficients is 0.7035, which less than 0.8. As for the VIF test, the VIF value of each explanatory variable is less than 10 (Table 5), suggesting that there is no multicollinearity problems. The concrete expression of the Tobit model is as follows:

$${\text{ULUEE}}_{{{\text{it}}}} = {\text{a}}_{{{\text{it}}}} + \beta_{{1}} {\text{GI}}_{{{\text{it}}}} + \beta_{{1}} {\text{GIE}}_{{{\text{it}}}} + \beta_{{2}} {\text{EDL}}_{{{\text{it}}}} + \beta_{{3}} {\text{IS}}_{{{\text{it}}}} + \beta_{{4}} {\text{OPL}}_{{{\text{it}}}} + \beta_{{5}} {\text{GTPL}}_{{{\text{it}}}} + \beta_{{6}} {\text{ICL}}_{{{\text{it}}}} + \beta_{{7}} {\text{IL}}_{{{\text{it}}}} + {\text{e}}_{{\text{i}}}$$
(3)
Table 4 The correlation test.
Full size table
Table 5 The VIF test.
Full size table

Table 6 shows that The Government intervention in the economy had a negative impact on ULUEE in the PRDUA during the study period. From 2006 to 2021, the proportion of financial expenditure to GDP gradually increased from 7.9 to 12.8%4, meaning that government intervention in the Pearl River Delta region has gradually increased. The local governments in the Pearl River Delta region need to avoid the excessive intervention of the market economy, and optimize fiscal expenditures to stimulate the vitality of the market economy.

Table 6 Tobit regression results.
Full size table

There was significant positive impact of Economic development level on ULUEE. The PRDUA is one of the most prosperous urban agglomerations in China. This conclusion is in accord with Liu et al.77 and He et al.78 With the improvement of economic development level, the contradiction between supply and demand of the urban land gradually worsens, and the urban land resources becoming increasingly scarce. The municipal governments of the region should constantly raise the threshold of urban land supply, increase the output capacity of urban land.

The regression coefficient of Opening-up level is significantly positive, indicating that raising opening up level has a beneficial effect on ULUEE. However, this is not to advocate blindly expanding foreign trade exports or short-term economic benefits. We should reasonably integrate the trade structure, and reduce the import and export of high energy consumption and high emission products, and support trade exchanges of high-tech and environment friendly industries.

Infrastructure construction level have presents a notable negative correlation with ULUEE, which meeting our expectations. In recent years, the regional infrastructure have been relatively perfect. Further improving the infrastructure is essential. The PRDUA should speed up the intelligence construction of modern transport, industry and infrastructure.

Both Industrial structure and green technological progress level have a positive impact on ULUEE, but not apparently. In recent years, the PRDUA has achieved rapid progress of both the industrial structure and green innovation technology. And the industrial level has moved from the lower-end of the global industrial chain to the middle- and higher-end72,73, which has significant increase in green innovation output. In order to play the role of the industrial upgrading, it needs to strengthen the industrial policy guidance and support, and improve the market mechanisms and industrial supervision. The local governments should provide more tax preferences, innovation awards, and system guarantees for independent innovation of enterprises, and set up a market-oriented innovation system.

Discussion

It is worth noting that the spatial gap of ULUEE between urban agglomerations tended to decrease after 2017. This feature is quite similar with Kong et al.79 and Chen et al.80 , who found that there is a decreasing spatial disparity in ULUEE in the Urban Agglomerations of the Yellow River Basin in China in the same period. Chen et al.80 also finds the same conclusion in the Yangtze and Yellow rivers in China. This paper acknowledges certain limitations that warrant further research. First, due to data constraints, the analysis could not extend prior to 2006. Expanding the study period would provide more comprehensive insights. Additionally, this research focuses on cities as the primary units of analysis, without considering districts or counties. Future studies should explore these smaller administrative units for a more granular understanding. Moreover, the paper only focused on the land use eco-efficiency of administrative spaces, future work should focus on the that of specific urban land, such as manufacturing land, commercial land, transport land, and so on.

Conclusions and policy suggestions

This study applied the super EBM model with undesirable outputs to calculate ULUEE for nine cities in the PRDUA between 2006 and 2021. Using the Tobit model, we also examined the factors driving efficiency improvements. The key findings are as follows: (1) From 2006 to 2021, ULUEE in the PRDUA ranges between 0.724 and 0.782, with the highest point was in 2007 and the lowest point was in 2014. (2) Guangzhou, Shenzhen, and Foshan maintained relatively high ULUEE levels throughout the study period, while Zhuhai, Zhongshan, Jiangmen, and Dongguan exhibited lower efficiency. (3) Economic development and openness contributed positively to ULUEE, while government intervention and Infrastructure construction had an adverse effect.

Based on these findings, we propose the following recommendations to enhance ULUEE in the PRDUA: (1) Cities within the PRDUA should strengthen collaboration to develop cohesive urban agglomeration plans, fostering coordinated growth across urban and rural areas. Encourage industrial transformation and upgrading, and develope high-tech industries and green industries. Strengthen cooperation and exchanges with advanced international regions, and introduce advanced technology and management experience. The governments should improve the land management system and strengthen supervision of the land market. Optimize the layout of infrastructure construction, and improve the utilization efficiency and ecological benefits of infrastructure. (2) For large cities such as Guangzhou and Shenzhen, which exhibit higher ULUEE and face limited land availability, efforts should focus on urban renewal projects, particularly revitalizing underutilized land in older districts and exploring underground space development. Relocating low-intensity industries to satellite cities, while bolstering high-end services, modern commerce, and creative industries, will further optimize land use. We will increase investment in research and development of green technologies and promote innovation and transformation of green technologies. (3) Dongguan, Zhongshan, Zhuhai, and Foshan, being adjacent to regional economic hubs, should leverage their strategic locations by adjusting their industrial and land use structures. Prioritizing innovation-driven industries and creative sectors will enable more efficient land resource allocation. (4) Jiangmen, Huizhou, and Zhaoqing, which are geographically distant from the core of the Pearl River Delta, should focus on accommodating industrial transfers from key development areas and promoting labor-intensive industries. These cities must also manage surplus rural labor while preserving ecological buffer zones and agricultural land to enhance environmental quality. These cities need formulate green industrial policies, guide industries to become green and low-carbon industries, and promote industrial upgrading and ecological transformation.