Introduction

As a public product, water resources are excessively consumed in the process of meeting the needs of human survival and economic development (Wang et al. 2019c; Wu et al. 2018; Zhang et al. 2017; Shang et al. 2017; Eamen et al. 2020). The shortage of water resources has become a major bottleneck, restricting the sustainable development of the economy and society (Xu et al. 2018; Guo and Shen 2016; Zhou et al. 2018). Water shortage is a complex issue affected by multiple factors such as industrial water use structure, technological progress, economic level, population, and water resources utilization ratio (Sun et al. 2011; Yao et al. 2019; Chen et al. 2019; Li et al. 2020; Li et al. 2019b; Long et al. 2019). Among the driving factors, technological progress runs through the whole resource use process, such as resource collection and distribution technology, resource recycling technology, etc (Moro et al. 2018), thus improving resource utilization efficiency (Li et al. 2019a; Song et al. 2018). Therefore, technological progress is the key factor in solving water shortage problems.

However, the ways to achieve technological progress are diverse and complex, including innovation exploration and efficiency improvement (Narvekar and Jain 2006; Harris and Woolley 2009; Jalonen 2012). In terms of technological innovation exploration, by upgrading technical tools and knowledge transformation, the overall technology frontiers can be expanded, which represents the highest technological levels (Hyvärinen et al. 2020). Another way to realize technological progress is to optimize factor allocation. That is the so-called efficiency improvement, which means more actual outputs with the same inputs. Both innovation frontier advancement and management efficiency improvement positively affect water resource conservation (Zhang et al. 2012; Zhang et al. 2017; Costa et al. 2007; Tuqan et al. 2020; Wu et al. 2018). On the one hand, water management policies drive water-intensive industries out of water-scarce regions (Chen et al. 2021; Wang et al. 2022). On the other hand, the implementation and transformation of innovative technologies are supported by financial subsidies and tax incentives (Li et al. 2019a; Wu et al. 2021). In comparison, there are uncertainties in long-term technological change, while the implementation and promotion of existing technology are more controllable. The government often plays an essential role in technological efficiency management (Zhang et al. 2017; Li et al. 2000; Yang et al. 2010; Zheng et al. 2014).

China is a water-scarce country, with a low per capita water resources share, and is listed as 1 of 13 water-poor countries by the United Nations. In this case, water resources have a restrictive effect on economic development, and for different regional groupings, there are still differences in their influence features (Gu et al. 2017; Denooyer et al. 2016; Voltz and Grischeck 2018). Given China’s acute water scarcity, extant studies have predominantly examined how technological progress mitigates water consumption, while largely overlooking the inherent limitations imposed by technological boundaries. Hence, it is necessary to consider technological innovation exploration or efficiency improvement separately. According to this, this article nests the Production-theoretical Decomposition Analysis (PDA) into the LMDI decomposition model to explore the key influencing factors of water consumption from the perspective of technological boundaries in China, considering the spatial and temporal differences.

Literature review

Amidst growing water scarcity crises, investigating the determinants of water conservation is critical for achieving sustainable development goals. Existing empirical studies have explored the influencing effects of various factors on changes in water consumption and water resource utilization efficiency, such as economic development, industrial structure adjustment, population growth, and technological progress (Cheng et al. 2023; Li et al. 2024a, b; He et al. 2025). Among these factors, economic development and population growth are generally considered the main reasons for the increase in water consumption, while technological progress is a key influencing factor for saving water. For example, Zhou et al. (2020) decomposed the changes in water consumption in China into 14 socio-economic influencing factors, and pointed out that adopting efficient irrigation and industrial water cycle technologies is the main reason for the slowdown in water consumption. Gong et al. (2025) used the LMDI to decompose the influencing factors of industrial water use in China’s Yangtze River Economic Belt, and the results showed that technological effect is the main force that restrains industrial water use. Similarly, Wang et al. (2019a), Deng and Zhang (2023) also confirmed the importance of technological progress in saving water.

Although the contribution of technological progress to water conservation has been widely recognized, Yang et al. (2022) further pointed out that technological progress has potential trade-off effects on water saving. The complex and non-linear relationship between technological advancement and water use efficiency necessitates a systematic analysis of its underlying conservation mechanisms. The ways to achieve technological progress are not singular, but previous research often confuses these. One critical pathway is pioneering innovation, which extends the aggregate technological frontier through knowledge transformation and technological upgrading (Wang and Feng 2020). This process effectively elevates production ceilings, enabling higher theoretical maximum output from equivalent input factors. Another one is efficiency management, which refers to the optimization of resource allocation to achieve more actual output with the same input at the existing technological levels (Narvekar and Jain 2006; Harris and Woolley 2009; Wang et al. 2019b; Liu et al. 2019). These two ways generally be mentioned in studies such as green total factor productivity (Ma et al. 2025) and carbon emission efficiency (Cheng et al. 2018; Wang et al. 2025), but it has not been fully utilized in water use efficiency research. Current research exhibits a notable gap in systematically incorporating both pioneering innovation and efficiency management within a unified analytical framework to quantify and compare their respective contributions to aggregate water consumption.

Our marginal contributions can be concluded in the following three aspects: a) An analytical framework for the macroscopic effect of technological progress on water consumption is built in this research from innovation exploration and efficiency improvement. Innovation exploration refers to advancing the technological boundary, explained by the technological change effect. Efficiency improvement refers to implementing and promoting technology at a certain technical level, explained by the technological management efficiency effect. However, current research always confuses the two. b) To quantify and identify technological change effect and technological management efficiency, we nested the Logarithmic Mean Divisia Index with Production-theoretical Decomposition to analyze how technological progress affects water consumption. It makes clarifying the two aspects of technological progress in different regions easier. c) As a case area, there are considerable differences in the economic development, technological progress, and technological management level in different regions of China, so analyzing the effect of technological progress on water resources consumption in the different regions of China from the perspective of spatial-temporal differences has practical value. It is helpful to put forward targeted policy recommendations based on regional resource endowments and development planning.

Methodology framework and data

Index Decomposition Analysis (IDA) have been widely used in quantitative analysis of influencing factors such as resource consumption and pollutant discharge in recent years, because of their flexibility and low data requirements, among which the LMDI method is the most widely used (Fernández et al. 2014; Ma et al. 2018; Zhang et al. 2019b; Zhao et al. 2025). The LMDI model can identify the overall contribution of technological progress to changes in water consumption, but it is unable to distinguish further the independent contributions of pioneering innovation and efficiency management (Narvekar and Jain 2006; Harris and Woolley 2009; Jalonen 2012). In contrast, the DEA model can measure technical efficiency and become an essential part of the PDA model, which can construct a decomposition framework for technological progress (Li 2010, Kim and Kim 2012). However, it cannot quantify the contributions of influencing factors to changes in water consumption. To accurately identify the contribution of pioneering innovation and efficiency management and clarify the impact mechanism of technological progress on water conservation, we construct the LMDI-PDA decomposition model, integrating the advantages of the two traditional decomposition models.

Data envelopment analysis model

When discussing the effect of technological progress on water consumption from innovation exploration and efficiency improvement, the boundaries of technological progress should be considered. Therefore, we apply the Data Envelopment Analysis (DEA). DEA originated from Farrell (1957), which focused on production efficiency. It is a relative validity analysis method based on multi-input and multi-output data. It can construct the most effective system based on input and output data, measure the distance between a single decision unit and the production frontier, and then measure the relative technical efficiency of a technology unit, which is also an important part of Li (2010), Kim and Kim (2012) introduced the PDA model into the framework of carbon emission decomposition to evaluate the carbon dioxide emission efficiency of a decision-making unit. The PDA method can decompose the effects of technological efficiency and technological changes in measuring production technology and energy use technology. In this paper, water resources input (W), capital stock (K), and the labor force (L) are regarded as inputs, and industrial added value (Y) as output in the production process, then the production possibility set of the region i with time series can be expressed as follows:

$${S}^{t}=\left\{({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t}):({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t}){{\Rightarrow }}{Y}_{i}^{t}\right\}$$
(1)

This research assumes that St is a closed convex set (Ruggiero 2007; Zhang and Cui 2016). This is because production theory indicates that limited input can only produce limited output. Then the distance between the relevant input and the optimal production frontier of region i in the t period can be expressed as follows:

$${D}_{{wi}}^{t}=\left({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t}\right)=\sup \left\{{{{\lambda }}}_{i}:({W}_{i}^{t}/{{{\lambda }}}_{i},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})\in {S}^{t}\right\}$$
(2)

The formula \({D}_{{wi}}^{t}\left({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t}\right)\) represents the distance between the actual input and the optimal production frontier (Cook and Zhu 2006; Bi et al. 2014). For the input of water resources, under the given output level, the less input, the better, which means \({D}_{{wi}}^{t}\left({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t}\right)=1\). But when the value of \({D}_{{wi}}^{t}\left({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t}\right)\) is less than 1, it means that the input of the water resources of the decision-making unit has not reached the optimal production frontier, and the efficiency is low. St represents a set of production technologies based on constant returns to scale. To further clarify the impact of technological progress on water use, we need to calculate four input distance functions: current distance functions \({D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})\), \({D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})\) and mixed-period distance functions \({D}_{wi}^{t}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})\) and \({D}_{wi}^{t-1}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})\).

LMDI-PDA decomposition model

Based on the measurement results of DEA model, we introduced the decomposition framework of PDA model into LMDI model and constructed the LMDI-PDA decomposition model to explore the driving factors of water consumption in China. Concerning Kaya’s identities and the expansion formulas of Johan et al. (2002), which analyze the impact mechanisms of various factors which are based on environmental sociology theory and environmental economic theory, the total water consumption can be decomposed as follows:

$$TW=\mathop{\sum}\limits_{i}\frac{{W}_{i}}{{V}_{i}}\cdot \frac{{V}_{i}}{{P}_{i}}\cdot \frac{{P}_{i}}{P}\cdot P$$
(3)

In the formula, Wi is the water consumption of region i, Vi is the industrial added value of region i, Pi is the population of region i, and P is the total population of the whole country.

Referring to the PDA decomposition method proposed by Zhou and Ang (2008), we use the geometric mean of stage t-1 or t to express the change in productivity which can avoid arbitrariness. The model is constructed as follows:

$$\begin{array}{c}T{W}^{t-1}=\mathop{\sum}\limits _{i}\frac{{W}_{i}^{t-1}\cdot {[{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})\cdot {D}_{wi}^{t}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})]}^{\frac{1}{2}}}{{V}_{i}^{t-1}}\cdot \frac{{V}_{i}^{t-1}}{{P}_{i}^{t-1}}\cdot \frac{{P}_{i}^{t-1}}{{P}^{t-1}}\cdot {P}^{t-1}\cdot \\ \frac{1}{{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}\cdot {[\frac{{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}{{D}_{wi}^{t}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}]}^{\frac{1}{2}}\\ =\mathop{\sum}\limits _{i}IPW{E}_{i}^{t-1}\cdot I{E}_{i}^{t-1}\cdot RP{E}_{i}^{t-1}\cdot P{E}^{t-1}\cdot IWV{E}^{t-1}\cdot TWTA{E}_{i}^{t-1}\end{array}$$
(4)
$$\begin{array}{c}T{W}^{t}=\sum _{i}\frac{{W}_{i}^{t}\cdot {[{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})\cdot {D}_{wi}^{t-1}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})]}^{\frac{1}{2}}}{{V}_{i}^{t}}\cdot \frac{{V}_{i}^{t}}{{P}_{i}^{t}}\cdot \frac{{P}_{i}^{t}}{{P}^{t}}\cdot {P}^{t}\cdot \\ \frac{1}{{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}\cdot {\left[\frac{{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}{{D}_{wi}^{t-1}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}\right]}^{\frac{1}{2}}\\ =\mathop{\sum}\limits _{i}IPW{E}_{i}^{t}\cdot I{E}_{i}^{t}\cdot RP{E}_{i}^{t}\cdot P{E}^{t}\cdot IWV{E}^{t}\cdot TWTA{E}_{i}^{t}\end{array}$$
(5)

Among them, \(IPW{E}_{i}^{t-1}=\frac{{W}_{i}^{t-1}\cdot {[{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})\cdot {D}_{wi}^{t}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})]}^{\frac{1}{2}}}{{V}_{i}^{t-1}}\) represents the potential water resources consumption of region i in the period t-1 (\(IPW{E}_{i}^{t}=\frac{{W}_{i}^{t}\cdot {[{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})\cdot {D}_{wi}^{t-1}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})]}^{\frac{1}{2}}}{{V}_{i}^{t}}\) represents the potential water resources consumption of region i in the period t), which is the possible water resources intensity effect. Notably, this term is the change of potential water intensity rather than real. The potential water intensity means the actual water intensity if the inefficiency of the water usage technology has been explored as much as possible (Kim and Kim 2012). \({{IE}}_{i}=\frac{{V}_{i}^{t}}{{P}_{i}^{t}}\) represents the per capita gross domestic product (GDP) of region i, which is the income effect. \({{RPE}}_{i}=\frac{{P}_{i}}{P}\) represents the proportion of the regional population in the total population, which is the regional population effect of region i. \({PE}\)=P represents the total population of China in different periods, which is the population size effect. \(IWV{E}_{i}^{t-1}=\frac{1}{{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}\) represents the technological management efficiency effect of region i in period t-1 (\(IWV{E}_{i}^{t}=\frac{1}{{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}\) is the technological management efficiency effect of region i in period t). \(TWTA{E}_{i}^{t-1}={[\frac{{D}_{wi}^{t-1}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}{{D}_{wi}^{t}({W}_{i}^{t-1},{K}_{i}^{t-1},{L}_{i}^{t-1},{Y}_{i}^{t-1})}]}^{\frac{1}{2}}\) is the technological change effect of region i in period t-1 (\(TWTA{E}_{i}^{t}={[\frac{{D}_{wi}^{t}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}{{D}_{wi}^{t-1}({W}_{i}^{t},{K}_{i}^{t},{L}_{i}^{t},{Y}_{i}^{t})}]}^{\frac{1}{2}}\) is the technological change effect of region i in period t).

According to the decomposition model, we decomposed the change in water resources consumption into six effects: potential water resources intensity effect (IPWE), income effect (IE), regional population effect (RPE), population size effect (PE), technological management efficiency effect (IWVE) and technological change effect (TWTAE). The water resources consumption model is constructed as follows:

$${{TW}}^{s}=\sum _{i}{{IPWE}}_{i}^{s}\cdot {{IE}}_{i}^{s}\cdot {{RPE}}_{i}^{s}\cdot {{RP}}^{s}\cdot {{IWVE}}_{i}^{s}\cdot {{TWTAE}}_{i}^{s}$$
(6)

Among them, \(s\in \{t-1,t\}\).

The LMDI method is divided into the “additive model” and “multiplication model”. To explain the decomposition results more clearly, we chose the “additive model” for factor decomposition (Frédéric and Quirion 2015). The factors are decomposed as follows:

$$\begin{array}{c}\Delta {W}_{IPWE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{IPW{E}_{i}^{t}}{IPW{E}_{i}^{t-1}}\\ \Delta {W}_{IE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{I{E}_{i}^{t}}{I{E}_{i}^{t-1}}\\ \Delta {W}_{RPE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{RP{E}_{i}^{t}}{RP{E}_{i}^{t-1}}\\ \Delta {W}_{PE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{P{E}_{i}^{t}}{P{E}_{i}^{t-1}}\\ \Delta {W}_{IWVE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{IWV{E}_{i}^{t}}{IWV{E}_{i}^{t-1}}\\ \Delta {W}_{TWTAE}=\mathop{\sum}\limits_{i}\frac{{W}_{i}^{t}-{W}_{i}^{t-1}}{{\mathrm{ln}}({W}_{i}^{t})-\,{\mathrm{ln}}({W}_{i}^{t-1})}{\mathrm{ln}}\,\frac{TWTA{E}_{i}^{t}}{TWTA{E}_{i}^{t-1}}\end{array}$$
(7)

The above formulas \({\Delta W}_{{IPWE}},{\Delta W}_{{IE}},{\Delta W}_{{RPE}},{\Delta W}_{{PE}},{\Delta W}_{{IWVE}},{\Delta W}_{{TWTAE}},\) represent the potential water resources intensity effect, income effect, regional population effect, population size effect, technological management efficiency effect, and technological change effect.

Data sources and processing methods

In this paper, we selected data on water resource consumption, economic growth, population change, and labor force in 31 provinces from 2000 to 2019. The data was selected up to 2019 because the outbreak of COVID-19 pandemic led to unconventional and non-trend changes in global economy, technology and management, causing abnormal declines in water consumption (Yang et al. 2022; Zhao et al. 2022; Henrysson et al. 2024). In addition, the statistical caliber of some indicators has been adjusted after COVID-19. To ensure the accuracy and comparability of our findings and draw regular conclusions, the data after 2019 is not included, which is consistent with the processing method of other research (Abbas et al. 2024; Zhang et al. 2025). Due to the lack of data, this study does not include Hong Kong, Macao, and Taiwan in China. The data were from the China Statistical Yearbook (National Bureau of Statistics), China Water Resources Bulletin (Ministry of Water Resources of the People’s Republic of China), and provincial statistical yearbooks. To eliminate the influence of price factors, we converted the added value of industries in each province according to the constant price 2000. Since there was no available capital stock data in the yearbook, we adopted the international sustainable inventory method to calculate the capital stock from 2000 to 2019. However, due to the limitations of statistical data, it is challenging to adopt the standard method of the perpetual inventory method. We assumed that capital efficiency adopts the model of geometric decline. The concrete formulas are as follows:

$${K}_{i(t+1)}=\left(1-{\delta }_{{it}}\right){K}_{t}+{I}_{i(t+1)}$$
(8)

Among them, Kit and Ki(t+1) is the capital stock of periods t and t + 1. δit represents the depreciation rate of the capital stock of region i in period t, and Ii(t+1) is the actual total investment of province i in the t + 1 period. The data on the nominal gross investment of each region is the total amount of fixed assets formed each year. This data is the nominal value, so it also needs to be treated as a real gross investment using the fixed assets investment price index. The depreciation rate of capital stock in this paper was calculated as 9.6%, which was obtained by Zhang et al. (2004). The capital stock in the initial year was estimated by the following formula:

$${K}_{i2000}={I}_{i2000}/({g}_{i2000}+{\delta }_{i2000})$$
(9)

Here, gi2000 refers to the GDP growth rate of region i in 2000, which was calculated by subtracting 100 from the GDP index in 2000.

Results

Analysis of the overall decomposition effect of LMDI-PDA

Based on the data and formulas in Section “Literature review”, IPWE, IE, RPE, PE, IWVE, TWTAE, and total water consumption change of China from 2000 to 2019 were estimated, as shown in Fig. 1.

Fig. 1
Fig. 1
Full size image

Decomposition factor effect of water consumption change in China from 2000 to 2019.

The total water consumption increased from 2000 to 2001. However, water consumption declined considerably for two consecutive years after 2001, possibly related to the national water-saving policy orientation in 2001 and the National Water-saving Planning Outline in China. From 2003 to 2013, total water consumption showed a fluctuating upward trend with a growth rate of 15.98%. After 2013, the change in total water consumption showed a downward trend.

From the decomposition effect of each factor, IPWE and TWTAE in China are negative, indicating that the changes in the factors have an inhibitory effect on the increase of water consumption. However, IE, the cumulative effect of RPE, the cumulative effect of PE, and the cumulative effect of IWVE in China are all positive, indicating that the changes in these four factors have a pulling effect on the increase of water consumption in China.

From the perspective of the contribution degree of each decomposition factor, IPWE is the key factor in promoting water consumption decline, while the contribution of TWTAE is relatively small. IE in all periods is always positive and far higher than the absolute value of RPE, PE, and IWVE, which shows that the change in the per capita GDP is the most critical factor leading to the increase in water consumption.

Analysis of factors’ decomposition effects of LMDI-PDA

The impact of individual driving factors on water consumption patterns exhibits significant temporal and spatial heterogeneity. Figure 2 presents the geographical distribution of the driving effects impacts for each factor and total water consumption.

Fig. 2: Distribution of driving factors in China’s Provinces from 2000 to 2019.
Fig. 2: Distribution of driving factors in China’s Provinces from 2000 to 2019.
Full size image

Changes in water consumption driven by potential water resources intensity effect (IPWE) (a), income effect (IE) (b), regional population effect (RPE) (c), technological management efficiency effect (IWVE) (d), technological change effect (TWTAE) (e). f Changes of total water consumption.

Potential water resources intensity effect (IPWE)

Suboptimal technological efficiency in water resource utilization leads to observed regional water use intensities consistently exceeding the theoretically optimal level. This gap is expected to narrow with ongoing technological advancements. This study proposes the novel concept of “potential water intensity,” which precisely captures the theoretically optimal utilization efficiency of water resources. Specifically, it represents the water use intensity when technological efficiency reaches the production frontier, corresponding to the minimum possible water consumption per unit GDP output.

The potential water resources intensity showed a gradual downward trend from 2000 to 2019, indicating that the water resources consumption per unit GDP decreased yearly (Fig. 1). From the point of the decline rate of the potential water resources intensity, the rate has slowed since 2010, indicating that the minimum water use per unit GDP has little potential to continue to decline, and the inhibition of water consumption has a weakening trend.

According to Fig. 3, IPWE has increased or decreased in China’s eastern, middle, and western regions. The absolute value of IPWE in the eastern region is the smallest, accounting only for 4.18% and 4.05% of that in the middle and western regions. At the same time, the absolute value of IPWE in the east region decreased more than that in the middle and west regions, indicating that the inhibition of IPWE on water consumption in the east is weakened and its reduction speed is fast. Additionally, the differences among provinces within the eastern region are the least, which indicates that the eastern provinces contribute similarly to IPWE. Generally speaking, IPWE has a strong restraining effect on the increase in water consumption in China, but its contribution level is declining continuously, especially in the eastern region. The impact of reducing water consumption per unit GDP on further reducing total water consumption will be limited in the future. However, in comparison, there is still room for reduction in the middle and western regions, and the gap between the central and western regions can be further narrowed.

Fig. 3
Fig. 3
Full size image

The IPWE, IWVE, and TWTAE in the East, Middle, and West of China.

Technological change effect (TWTAE)

TWTAE measures how innovations from scientific discoveries and technological inventions improve water efficiency by capturing the change of the optimal production frontier boundary of each decision-making unit. According to data from Chinese Research Data Services Platform (CNRDS), from 2000 to 2019, cumulative patent authorizations in China’s water production and supply industry rose from 90 to 31720, increasing by 352 times. Key technological changes in water saving cover multiple aspects, such as agriculture, industry, and daily life (Zhang et al. 2019a; Zhou et al. 2020). During agricultural production, water-saving irrigation technologies, such as drip irrigation, sprinkler irrigation, have been promoted and applied. In industrial production, water recycling systems have been introduced. In daily life, the invention and popularization of water-saving appliances play an important role. These technologies reduce water losses and improve water production efficiency. From 2000 to 2019, TWTAE was negative, which shows that TWTAE has a certain inhibitory effect on water consumption in China. As shown in Fig. 3, the TWTAE in the eastern region is always negative, while the TWTAE in the middle and western regions is positive in most years. Furthermore, the absolute value of the cumulative negative effect in the eastern region brought by technological change is about 7.62 and 13.28 times that in the middle and western regions respectively. And the internal differences in the east, the middle, and the west regions are diminishing. This suggests that the production frontier in eastern China advanced faster than in central and western regions, while showing gradual convergence across all internal regions.

To further analyze the progress of the production frontier in the east, middle and west of China, we calculate the ratio of the water use efficiency of period t to period t-1 with the factor input in the t-1. If the ratio is less than 1, the production frontier improves compared with the previous period, and vice versa. As shown in Fig. 4, the average ratio of the eastern region is less than 1, while the ratio of the middle and western regions is greater than 1. This shows that the production frontier in the eastern China keeps advancing, while the production frontier in the middle and western regions did not advance smoothly.

Fig. 4
Fig. 4
Full size image

The technology gap ratios and the change of the optimal production frontier boundary in the East, Middle, and West of China.

Differences in the development model and resource endowments between the east, middle, and west of China may shape this significant gap. We use the technology gap ratio (TGR) to explain this difference. TGR refers to the ratio of the green efficiency value of water resources under the joint frontier to that under the group frontier. The closer the TGR is to 1, the closer the group’s production level is to the joint production technology level, and vice versa, the farther the horizontal distance is from the joint technology production level. As shown in Fig. 4, as the pioneer of China’s reform and opening-up, the eastern region benefits from advanced economic development and greater emphasis on technology adoption and innovation, leading to superior water-use efficiency. Consequently, its water-use technology has converged toward the joint production frontier. However, the water use technology in the middle and western regions only reached 83.36% and 86.33% of the joint frontier respectively. This is limited by the local resource allocation method and industrial structure, and there is a relatively large technological gap relative to the production frontier.

Technological management efficiency effect (IWVE)

IWVE reflects changes in water consumption caused by differences in technological management efficiency in different regions under the same technical level. According to water optimal allocation theory, implementing appropriate management approaches can improve water use patterns and efficiency. Improving technological management efficiency can enable production units to catch up with the highest technical levels, thereby achieving optimized allocation and conservation of water resources. As shown in Fig. 3, from 2000 to 2019, IWVE in China was positive and negative at different time stages, and the cumulative effect was positive. This shows that IWVE promotes the increase of water consumption, and its contribution rate is 240.12%. From the point of time variation, IWVE in China has strong fluctuation, which can be divided into three stages: 1) from 2000 to 2004, the effect of technological management efficiency changed from positive to negative, indicating that IWVE on promoting the reduction of water resources consumption gradually weakened. This means that during the “Tenth Five-Year Plan” period, China considered the improvement of water use efficiency in the process of rapid economic progress. 2) From 2005 to 2014, IWVE in China increased gradually, indicating that the economic development model of some provinces with less water consumption began to change to the direction of low energy consumption and high efficiency during the “11th Five-Year Plan” and the “12th Five-Year Plan”, while provinces with more water consumption have made limited progress in improving water resources efficiency. Changes in ranking among provinces caused changes in IWVE. 3) After 2015, IWVE declined gradually, and even became negative, indicating that provinces with more water consumption improved technological management efficiency in this period. From a spatial perspective, Cumulative IWVE shows similar positive and negative features in the eastern, middle and western areas as TWTAE. But this cumulative negative effect in the eastern regions is much smaller than the positive one in the middle and western regions.

Income effect (IE)

Figure 1 shows that IE has been positive from 2000 to 2019, and its contribution to the increase in water consumption in China is the largest. This is because water resources are an important input factor to support China’s economic growth, and a higher level of economic development often means a larger demand for water resources. From the perspective of annual changes, IE shows a trend of rising first and then declining. This shift predominantly reflects China’s economic growth mode transition after the 2008 financial crisis, whereby the growth rate slowed down and IE gradually declined from its peak. This trend is consistent with the theory of water resources complex systems, which emphasizes bidirectional dynamics: while water resources provide fundamental inputs for socioeconomic development, economic and social activities actively reshape water consumption patterns.

From the perspective of regional differences, the contribution of diverse regions to China’s water consumption growth varies greatly. Guangdong contributed the most, while Beijing contributed the least, with a significant difference. This is mainly related to the economic development level in different regions. Population base and economic output are the key factors affecting the contribution of IE in different regions. As an important factor of production, economic growth characterized by industrialization and urbanization will inevitably be accompanied by an increase in the use of water resources.

Regional population effect (RPE)

The cumulative contribution of RPE in China is positive, with a contribution rate of 47.24%, which shows that RPE has a pulling effect on the increase in water consumption. As shown in Fig. 5, the cumulative effect of RPE was negative in 15 provinces and positive in 16 provinces from 2000 to 2019. We selected three provinces (Guangdong, Hubei, and Shandong) with representative regional population distributions for analysis. When the proportion of the population in the provinces decreases, RPE promotes the reduction of water consumption. The characteristics of regional population distribution are related to many factors. The most direct factors are the local fertility policy implementation and the population migration among provinces. At the same time, this is also linked to differences in the level of economic development among provinces in China.

Fig. 5
Fig. 5
Full size image

Regional population distribution effect of China’s Provinces.

Population size effect (PE)

The cumulative contribution of PE in China from 2000 to 2019 is positive, with a contribution rate of 228.13%. This shows that the change in the total population obviously promotes the increase of water consumption in China. China’s total population grew by 11.42% during this period. Except for the decline in the total population from 2004 to 2005, the growth trend remained in the remaining years. Considering the change in PE, the increase in PE will significantly promote the increase in water consumption in China. This can be summarized into two main reasons: 1) Water resources are an indispensable and vital resource for human survival. The increase in population will inevitably be accompanied by an increase in water consumption (Brelsford and Abbott 2017). 2) The increase in population size will inevitably lead to the expansion of economic scale and ultimately to the further growth of water consumption (Jiang and Li 2016). With the continuous liberalization of China’s “Third child” policy, China’s population size may grow relatively rapidly in the future, which will also pressure China’s water consumption.

Discussion on water resources conservation path

The research on the impact of technological progress on water resource consumption is already abundant, proving the positive contribution of technological progress (Zhou et al. 2020; Gong et al. 2025). Relatively, we have built an analytical framework that further decomposes technological progress, which can provide a richer perspective for improving water utilization efficiency from two aspects: pioneering innovation and efficiency management. Results show that innovation exploration can save water, while the management efficiency effect increases water consumption. Based on this, regional water use strategies are proposed. On the one hand, to enhance the positive contribution of pioneering innovation, the government should provide necessary innovation elements and fully leverage market incentives. On the other hand, regional efficiency management can be improved through government performance evaluation, property rights protection, and advanced technology promotion. These measures can reduce the negative impact of efficiency management, which may have been overlooked in other studies.

Innovation advance: water intensity reduction

Given the converging marginal effects of water-saving technologies and inherent uncertainties in technological advancement, fostering robust innovation platforms and pushing the water-efficiency frontier remain critical priorities. According to the analysis of IPWE and TWTAE, reducing water resource consumption per unit GDP and developing water resource technology are critical factors in restraining the increase in water consumption. The contributions of these influencing factors are similar to existing research findings (Long et al. 2019; Wang and Wang 2020; Du et al. 2022; Gong et al. 2025). However, there is a downward trend in the contribution of IPWE and TWTAE to the reduction of water consumption. This means that the upgrading speed of water resources utilization technology in China has slowed down, and there is still much room for improvement. However, the expansion of technological frontiers is inherently uncertain, requiring substantial investment in innovation factors and robust science policy support. The government should provide the necessary elements of technological innovation in water resources, such as policy guidance, relevant talents, and funds for supporting innovative enterprises. In this way, innovation carriers such as scientific and technological innovation industrial parks, scientific and technological talent platforms, and high-tech enterprise cultivation libraries can be formed. On the other hand, it is necessary to leverage market incentives. For example, water resources taxation policies can incentivize technological innovation that enhances corporate profitability, fostering market-driven competition and enabling sustainable upgrading of water-use technologies.

Management improvement: water-saving efficiency increase

The efficiency of technological management plays an important role in water resources use. In particular, more attention should be paid to water-intensive provinces. Under the constraints of technology boundaries, the IWVE of different provinces varies greatly, which is caused by the gap in management level between provinces. As evidenced by Qiao et al. (2020), in northwest China, the level of water technology is not the key factor in eliminating water resource limitations. At the current level of technology, improving water technological efficiency is crucial for alleviating water resource shortages. The difference in water-saving technology management mechanisms and resource allocation models in each province will affect the water resources utilization efficiency in different provinces. Therefore, first, water resources management should be included in government performance assessment to fully mobilize the enthusiasm and effectiveness of local government and promote the inclination of talents, funds, and other factors. Second, the government should protect property rights and regulate the technology market, to support the implementation of green technology innovation. In addition, the government should guide the promotion of advanced technologies, such as setting up demonstration zones for high-tech industries for technology demonstration and exchange. Third, from the perspective of the overall policy layout, strengthening the management and control of water resource utilization in water-intensive provinces and implementing some pilot policies will contribute to the conservation of water resources.

Technological gap and other differences among regions

The technology gap between the central and western regions and the eastern regions is still huge. It’s important to make up the technology gap. The level of development and the modes of economic development in the east of, middle, and western regions of China are quite different. The middle and western regions of China continue to receive water resources-intensive enterprises from the eastern region, resulting in greater pressure on water resources. Meanwhile, limited by the economic foundation and brain drain in the middle and western regions, they received industrial transfer. Still, they did not make up for the technological gap. Therefore, the government should build a differentiated evaluation system for development goals for water resource utilization on the premise of combining efficiency and fairness. It is necessary to transform traditional economic compensation into technical compensation. The government can implement the paired assistance policy to encourage the central and western regions to introduce advanced water-saving technologies from the east. Additionally, reasonable population flow is conducive to water conservation, so the government should carry out water-saving planning according to regional conditions. Based on the analysis of the effect of RPE, water-saving technology and resource endowment in eastern China are relatively advanced. Interregional population migration from China’s water-stressed inland areas to the technologically advanced eastern region enhances water conservation. Guiding the population to transfer to cities with high water resources carrying capacity is conducive to improving the ecological carrying capacity and relieving the production and living functions of areas with weak water resources carrying capacity. Therefore, the government should build an orderly passage of population flow from the central and western regions to the eastern regions. It is worth noting that population mobility must meet the actual needs of social production, otherwise it may have a more negative impact on economic development. Designing population development goals concerning water resources carrying capacity is necessary.

Conclusion

The shortage of water resources is a critical problem restricting China’s sustainable development. We analyze the potential water resources intensity effect, technological management efficiency effect, and technological change effects from the perspective of the technological progress frontier using a nested model of the Logarithmic Mean Divisia Index and Production-theoretical Analysis (LMDI-PDA). Based on the effects of various factors on total water consumption in China, more specific conclusions and policy recommendations were put forward.

Firstly, during the research period, the potential water resources intensity effect and the technological change effects were the main factors in cutting down water consumption, while the technological management efficiency effect increased the water consumption. The potential water resources intensity effect made the most significant contribution to reducing water consumption in China from 2000 to 2019, reaching 8078.41 billion m3, followed by the technological change effect. In contrast, the technological management efficiency effect promotes increased water consumption, and its contribution rate is 240.12%. Therefore, the design of the water-saving path in China should focus on cultivating water resource innovation carriers and accelerating the advancement of water-saving technological boundaries. And it’s also important to improve the efficiency of technological management.

Secondly, there are also distinct differences in the three driving factors among China’s eastern, middle, and western provinces. Specifically, the eastern regions have the smallest potential water resources intensity effect, accounting only for 4.18% and 4.05% of that in the middle and western regions. The technological change effect in the eastern regions is always negative, while that in the middle and western regions is positive in most years. The technological management efficiency effect shares similar positive and negative features of the eastern, middle and western areas with the technological change effect. But this cumulative negative effect in the eastern regions is much smaller than the positive ones of the middle and western regions. Based on these, the government should focus on making up the technology gap between the middle and western regions and the eastern region. Therefore, the government should establish a differentiated evaluation system for development goals. It is necessary to encourage the middle and western regions to introduce advanced water-saving technologies from the east.

Besides, the cumulative contribution of the income effect, regional population effect, and population size effect to increase China’s water consumption is all positive, among which the income effect ranks the top. From the spatial perspective, the three effects all show significant regional heterogeneity. Given the income effect, Guangdong contributed the most, while Beijing contributed the least. And 15 provinces represent negative cumulative regional population effects and 16 are positive from 2000 to 2019. Reasonable population flow is conducive to water conservation, so the government should tailored water-saving planning based on regional conditions.

Notably, although this study explores water resource utilization from the perspective of the efficiency frontier and draws more concrete and feasible conclusions and suggestions than other studies, there are still some limitations. The study focuses on the inter-provincial level but does not measure the impact of the sub-industrial structure and industrial development level on China’s water consumption. In future research, measuring the impact of the industrial structure on water consumption can provide more detailed information about the industrial structure.