The impact of water scarcity on energy poverty and its mechanism: evidence from a rural household analysis in northern China Open Access

The United Nations Sustainable Development Goals (SDGs) call for addressing water scarcity (WS; SDG6) and eliminating energy poverty (EP; SDG7). This study explores the link between these two issues to offer insights for sustainable development. WS, defined as insufficient water resources to meet human production and daily life needs, has become a pressing issue due to climate change and economic development (Daghagh Yazd et al., 2020). It threatens agricultural productivity, exacerbates environmental degradation, and increases the risk of social and political instability (Oswald Spring, 2011; Mekonnen & Hoekstra, 2016; De Almeida Castro et al., 2018; Ikuemonisan et al., 2021; Castelvecchi, 2023).

In areas where agricultural production relies on groundwater, WS can lead to higher energy consumption for irrigation, driving up costs (Qiu et al., 2018; Shi et al., 2022a). This makes rural households more vulnerable and could be a factor in EP. EP refers to the inability to access affordable, reliable, and modern energy services, significantly hindering household well-being (Awaworyi Churchill et al., 2022). It is characterized by unaffordability, where a large share of income is spent on energy, and inaccessibility, where energy services are insufficient or unavailable (Lin & Wang, 2020; Lin & Zhao, 2021; Nie et al., 2021; Shi et al., 2022a, b). EP has broad socio-economic impacts, affecting health and reducing opportunities for development (Samarakoon, 2019; Li et al., 2022).

Northern China provides an ideal case study for examining the relationship between WS and EP. This region faces extreme WS, with only 19% of China's water resources, yet it supports 65% of the country's arable land and 50% of its agricultural production. The phenomenon of EP in the context of WS is not only intriguing but also a critical and captivating issue for sustainable development. The intersection of WS and EP presents complex challenges that warrant further investigation. However, the current literature lacks a comprehensive analysis of the interaction between WS and EP from the perspective of agricultural production. Given this backdrop, our study aims to address this research gap by investigating the mechanisms through which WS contributes to EP. This study identifies two key mechanisms: 1. Substitution effect: Farmers may increase fertilizer use to compensate for WS, leading to higher production costs and reduced income. 2. Labor loss effect: Increased time spent fetching water reduces labor available for off-farm employment, deepening EP (Booker & Trees, 2020).WS not only increases the direct cost of energy for irrigation but also worsens EP through changes in agricultural practices and labor dynamics (Zhang et al., 2019; Zarei et al., 2020; Pérez-Blanco & Sapino, 2022). This study aims to clarify these mechanisms and contributes to policies addressing the dual challenges of WS and EP.

Northern China makes an excellent case study for investigating the link between WS and EP (Qiu et al., 2018; Wang et al., 2020a, b; Shi et al., 2022b). This region faces extreme WS, accounting for only 19% of China's water resources while supporting 65% of the country's arable land and 50% of agricultural production (Song et al., 2018; Wang et al., 2020a, b). The phenomenon of EP in the context of WS is not only fascinating but also a critical and compelling issue for sustainable development. The intersection of WS and EP raises complex issues that require further investigation.

There is already a substantial body of research discussing the relationship between WS and agriculture (Liu et al., 2017; Perrone, 2020; Layani et al., 2022; Leal Filho et al., 2022; Ingrao et al., 2023; Levintal et al., 2023). However, the current literature does not provide a comprehensive analysis of the interaction between WS and EP in terms of agricultural production. Given this context, our study seeks to fill a research gap by examining the mechanisms by which WS contributes to EP. This study identifies two main mechanisms: Substitution effect: farmers may use more fertilizer to compensate for reduced agricultural yields due to WS, resulting in higher production costs and lower income (Gallic & Vermandel, 2020; Aragón et al., 2021). Labor loss effect: Spending more time obtaining water, due to queues, waiting, and disagreements, reduces the labor force available for non-farm employment, thereby contributing to EP (Booker & Trees, 2020). This research seeks to clarify these mechanisms and contribute to policies addressing the dual challenges of WS and EP. China is the largest developing nation, and the findings of the study conducted in China may be helpful to other regions experiencing WS and EP.

Due to the impacts of climate change and increasing demand, WS has become a critical issue for its socioeconomic effects, particularly concerning agricultural production and energy consumption in rural areas (Qiu et al., 2018; Balasubramanya et al., 2022; Jones et al., 2024; Marbler, 2024; Rathore et al., 2024). However, a thorough review of the current literature reveals two major shortcomings.

First, the metrics used to measure WS are often inadequate. WS is frequently assessed through subjective perceptions using self-reported household methods or economic perspectives focusing on the sustainable availability of low-cost water (Atikul Islam et al., 2013; Singh et al., 2018; Yoon et al., 2019; Aguilar et al., 2022). These methods pose significant challenges when analyzing the socioeconomic impacts over larger regions and extended periods. The variability in subjective inquiry methods across different regions results in a lack of unified standards, making it difficult to accurately measure WS on a larger scale and to assess its related impacts precisely. Moreover, existing research rarely utilizes objective, satellite-derived data to measure regional WS. Studies have often used WS as a study context, conducting case studies without selecting specific indicators or methodologies to precisely measure WS (Booker & Trees, 2020). Utilizing more scientific and objective satellite-derived data would facilitate broader and more extended measurements of WS, providing a solid foundation for more representative and in-depth analyses (Unfried et al., 2022).

Second, the literature primarily focuses on the impacts of WS on production, labor allocation, health, and farm household livelihoods (Atikul Islam et al., 2013; Singh et al., 2018; Yoon et al., 2019; Aguilar et al., 2022). While some research has explored strategies for managing WS, such as implementing effective early warning systems (Muyambo et al., 2024), existing studies largely overlook the relationship between WS and EP. WS can directly increase the cost of water extraction, often requiring more energy to pump water from deeper levels or using more powerful pumps to meet irrigation demands (Qiu et al., 2018; Balasubramanya et al., 2022). The increase in energy consumption directly affects the household energy budget, exacerbating EP. Additionally, indirect mechanisms, such as increased fertilizer use to compensate for reduced water availability (Aragón et al., 2021), lead to higher agricultural production costs and a disguised reduction in household income. Moreover, WS often necessitates increased labor for water fetching and irrigation, reducing time for off-farm employment and thereby increasing the risk of EP (Booker & Trees, 2020). Although part of the literature explores the potential socioeconomic impacts of WS in rural areas, it does not deeply analyze the effects and mechanisms of WS on EP.

Currently, there is no unified definition of EP in the academic community, both domestically and internationally (Guevara et al., 2023; Lu & Ren, 2023; Wang & Du, 2024). At the micro level, EP is typically considered at the household level, focusing on energy availability and affordability (Lin & Zhao, 2021). This can be divided into subjective perception and objective measurement, with objective indicators being more accurate and reflecting the real situation better (Cheng et al., 2021). Objective measures provide precise data that are essential for policy-making (Charlier & Legendre, 2019), and many scholars have adopted such metrics in their studies.

Lewis (1982) defined EP as the inability of households to afford energy to maintain a warm living environment. Boardman (1991) defined it as households spending more than 10% of their income on energy, which was approximately double the median energy expenditure share among UK households at that time (Liddell et al., 2012). Moore (2012) also used the threshold of twice the median energy expenditure share to define EP. Since the 10% threshold could overestimate EP by including high-income households, some scholars have revised it to consider only low-income households below the third income decile (Kahouli, 2020). Recently, the ‘low-income high cost’ (LIHC) measure by Hills (2011) has been widely adopted. This relative measure excludes high-income, high-energy-consuming households, focusing instead on those whose energy expenditure exceeds the median for their area but whose per capita income is below 50% of the area median. This relative measure excludes the interference of high-income and high-energy-consuming households. This measure defines EP by the following criteria: if their energy expenditure exceeds the median household energy expenditure for their area and year, but their household income per capita is below 50% of the median value for their area. This method has proven effective in measuring EP in China (Lin & Zhao, 2021; Nie et al., 2021; Shi et al., 2022b). At the macro level, scholars have also studied EP. Awaworyi Churchill & Smyth (2020) utilized multiple metrics to explore EP in-depth. Wang et al. (2015) developed an EP index using four categories and nine variables to analyze its status in China, revealing changes from 2000 to 2011. Zhao et al. (2021) introduced new variables to calculate China's EP indicators and assess various impacts (Dong et al., 2021). The literature outlines three main characteristics of EP: meeting basic energy needs, energy service quality, and accessibility. Ensuring minimum energy costs for survival activities is a key measure (Barnes et al., 2011; Chakravarty & Tavoni, 2013). The quality of energy services links EP with economic poverty (Hills, 2011), and indicators like traditional biomass use and electricity access reflect energy service availability (Pachauri et al., 2004).

Numerous studies have investigated the factors that contribute to EP, such as flooding, energy efficiency, income level, building conditions, non-agricultural employment, carbon tax, and so on (Sadath & Acharya, 2017; Winkler, 2017; Liu et al., 2018; Qu & Hao, 2022; Kyprianou et al., 2023; Okyere et al., 2023). Low energy efficiency, high household energy expenditure, and low income are significant causes of EP (Ürge-Vorsatz & Tirado Herrero, 2012; Ullah et al., 2024; Zhang et al., 2024). The issue of EP in rural areas is a multifaceted phenomenon that has garnered significant attention from scholars, leading to extensive and comprehensive research endeavors. A study conducted by Gafa & Egbendewe (2021) in rural Senegal and rural Togo revealed that factors such as low income, gender of the head of household, household size, and composition, and the effort exerted in collecting fuel (measured by distance traveled) were identified as the primary determinants of EP. Gafa et al. (2022) demonstrate that the presence of opportunity costs in energy collection is a contributing factor to the issue of EP. Cyrek & Cyrek (2022) highlight the existence of a rural–urban divide that potentially influences the prevalence of EP in rural areas. Furthermore, the variables of democracy and governance have an impact on the issue of rural EP (Acheampong et al., 2023). The variations in crop types cultivated have an impact on the issue of EP (Ahmed & Gasparatos, 2020). There are discernible disparities in EP between genders in rural areas of Bangladesh (Moniruzzaman & Day, 2020). The influence of energy structural transformation and governance patterns on rural households' EP is of significant importance (Yadav et al., 2019). Previous research has also investigated the influence of socio-cultural background and social interaction on the phenomenon of rural EP (Kumar, 2020; Li & Ma, 2023).

EP does not completely overlap with income poverty and represents a form of material deprivation. Economic downturns intensify EP, disproportionately affecting vulnerable groups, such as retirees and women living alone (Costa-Campi et al., 2024). Recent studies have analyzed the impact of unexpected energy price surges, such as those from the Russia–Ukraine conflict, on household EP (Nadimi et al., 2024). Combining fiscal and behavioral interventions may help achieve climate change and EP alleviation goals (Della Valle et al., 2024). Improving energy efficiency is critical for reducing EP (Streimikiene et al., 2020). EP partly arises from households' inability to afford energy, creating a vicious cycle of low productivity and limited energy access. In rural Qinghai-Tibet, households often compromise health and comfort by using cheap biomass and coal to save on energy costs (Liu et al., 2018). Collecting firewood consumes significant labor, affecting education and economic activities, thus reducing long-term income potential. Building conditions also influence EP, and solar houses are a vital solution (Liu et al., 2018). Access to modern energy reduces the time and effort required for energy collection, improving economic and social status. Breaking the cycle of EP requires improving access to modern energy services. Modern energy enhances productivity and frees up time for income-generating activities, such as non-agricultural employment, thus increasing income (Shi et al., 2022a, b). In the process of eliminating rural EP, China faces several challenges, including limited energy infrastructure, a large population living in poverty, serious population ageing, and limited environmental awareness (He et al., 2018).

While existing research is increasingly examining the environmental impacts of EP, the focus has predominantly been on temperature rather than WS. A small number of studies have recently emerged that have analyzed the impact of abnormal temperature shocks on rural household EP (Feeny et al., 2021; Que et al., 2022; Li et al., 2023). It is therefore logical that studies investigating the impact of environmental factors (e.g. WS) on EP in rural households should prioritise the agricultural production component. Agricultural production is the stable livelihood and social security of last resort for smallholder farmers, who must therefore be committed to ensuring appropriate irrigation practices. For numerous smallholder farmers, the rationale for action is more inclined toward survival logic than economic logic. There is limited literature based on this perspective to examine the potential mechanisms by which EP impacts rural households.

After reviewing the current literature on WS and EP in the context of sustainable rural development, several critical gaps emerge. First, few scholars have empirically verified the impact of WS on EP among rural households. These issues are integral to the SDGs and have significant research implications. Second, there is a lack of large-scale, long-term studies using satellite data to measure regional WS and correlate it with household socioeconomic data. Existing studies on WS are often small-scale, short-term surveys based on subjective perceptions. Third, current studies struggle to explain the mechanisms through which WS impacts rural households' EP. An in-depth examination of these mechanisms can inform more effective and targeted interventions to help rural households in water-scarce areas reduce the risk of EP.

The purpose of this paper is to examine the potential impact of WS on rural household EP. To this end, we propose the following two research questions to address existing research gaps:

Research Question 1: What is the impact of WS on EP?

Research Question 2: What is the role of labor loss and factor substitution effects in the mechanism by which WS impacts EP?

Research Question 3: How do smallholder farmers' decisions, influenced by survival rationality, mediate the relationship between WS and EP?

Research Question 4: How do the spatial and temporal dynamics of WS correlate with variations in EP indicators across rural households?

Research Question 5: What policy interventions can effectively address the dual challenges of WS and EP, particularly in regions with limited resources and infrastructure?

These questions form the foundation of the study's objectives, aligning with its aim to bridge gaps in understanding the intersection of environmental challenges and socio-economic vulnerabilities. The primary contributions of this study are outlined below: First, the integration of satellite data with socio-economic survey data facilitates a more scientific analysis. Second, it addresses a deficiency in the current literature regarding the correlation between WS and EP. Third, it further examines the mechanistic pathways of the interaction between WS and EP, thereby aiding policymakers in acquiring a more profound understanding of the associated issues.

The creation of precise and scientifically valid datasets is essential for deriving accurate and credible conclusions. This study will conduct a comprehensive analysis of the dataset created by merging satellite data with socio-economic data. This paper employs satellite data equivalent to liquid water thickness data from the Gravity Recovery and Climate Experiment (GRACE) project. The data are derived from the push-pull effect on two NASA-launched GRAIL artificial satellites to determine gravitational variations at the Earth's surface, providing monthly data since their launch in 2002 and publicly available on the NASA EARTHDATA website. The resolution of this dataset is 1/2° latitude × 1/2° longitude (55 km × 55 km). Small variations in the distance between the two satellites' orbital paths as they orbit the Earth are used to measure changes in water mass. These small variations originate from changes in the Earth's gravity field. The smaller the mass of the Earth, the weaker the gravitational field, while the larger the mass, the greater the gravitational force. This measurement is based on the assumption that water is the material on Earth that experiences significant mass fluctuations (Cooley & Landerer, 2019). Water mass changes can be monitored through various factors, including irregularities in rainfall, accelerated water outflows due to evapotranspiration, excessive utilization of water resources, occurrences of groundwater droughts, and alterations in runoff. This can be achieved by utilizing a unique dataset known as GRACE, which enables the measurement of total water mass changes at the level of individual grid cells. Small changes in the distance between two satellites that track each other as they orbit the Earth are used to measure changes in water mass.¹ These minute changes are the result of variations in the gravitational field of the Earth. A weaker gravitational field on Earth arises from less mass, whereas a greater pull emerges from more mass. The measurement is based on the assumption that water mass is the only changing mass on Earth (Unfried et al., 2022). The technical details of how GRACE uses gravity to measure water mass are not the focus of this study. Please check the official GRACE website for details.²

Furthermore, our socioeconomic data are sourced from the China Family Panel Studies (CFPS). A nationally representative sample of Chinese homes is available via the CFPS data collection, which covers 25 provinces and 162 counties that together account for 95% of China's total population (Yang et al., 2023). The CFPS provides comprehensive socioeconomic and demographic information for households. CFPS data allow us to accurately measure the EP of rural households. To serve the study objectives, the sample was restricted to rural households in northern China. The study encompassed a total of 9,083 households, with 2,243 households surveyed in 2012, 2,163 households in 2014, 2,543 households in 2016, and 2,134 households in 2018. The descriptive statistical analysis of the spatial distribution of the core variables in Section 4.1 illustrates the geographic location of the study area.

Changes in the mass of water serve as a valuable indicator for determining the amount of water that is available in various regions (Unfried et al., 2022). Changes in the mass of water in a region effectively reflect the availability of water resources in that region. This investigation utilizes the method of calculating the annual change in water mass, defined as the difference between the water mass measurements taken in the current year and those taken in the previous year, to quantify WS. This strategy posits a connection between the degree of water mass fluctuations and the available water, with the reduction in available water serving as the essence of WS. Significantly lower water mass changes indicate a decrease in the available water resources. This suggests a scarcity of available water during that time period, making it impossible for the water mass to oscillate any further. This signifies an increased scarcity of water. A substantial enhancement in water mass change signifies that the region possesses ample water resources, implying an improvement in water availability and a decrease in WS. In this study, the negative value of annual water mass change serves as the primary indicator for accurately assessing the extent of existing WS. This transformation enables the study to establish a direct correlation between the extent of water mass change and the severity of WS (Unfried et al., 2022). A higher value indicates increased severity of WS. Lower values indicate reduced WS.

This method offers several advantages in assessing the degree of WS. Initially, it considers both natural and anthropogenic factors that affect water availability. These factors encompass fluctuations in precipitation, evapotranspiration rates, and anthropogenic water usage. This method incorporates various elements of the hydrologic cycle, such as surface water, soil moisture, and groundwater, to offer a holistic perspective of the water system. The methodology emphasizes alterations in overall water mass. Second, the ability to assess WS through changes in water mass facilitates the capacity to perform uniform analyses across diverse regions. This method mitigates the inaccuracies linked to various approaches that pose subjective inquiries regarding WS and assesses it based on a uniform standard applicable to all.

In order to assure the accuracy of measurements and the robustness of estimate findings, this research used the following four different indicators to quantify EP: (EP1–EP4). Double the median proportion of entire income – EP1 (Moore, 2012); 10% measure – EP2 (Boardman, 1991), showing energy expenditures in excess of 10% of household income; revised 10% measure – EP3 (Kahouli, 2020). This research adopts EP3 indicators that only include low-income households whose income is still below the third decile of the distribution of household income since the 10% measure may overestimate the incidence of EP because it includes higher-income families (Kahouli, 2020).

The low-income high-cost (LIHC-EP4) indicator was also developed as a consequence of current research to quantify EP. Households are considered EP by LIHC if their annual energy costs are more than the provincial and wave-specific medians, but their per-person income is less than half of the provincial and wave-specific medians (Hills, 2012). The effects of high-income, energy-intensive families are significantly reduced by the LIHC index. The LIHC indicator is more accurate since it eliminates any possible bias brought on by high-income households. The LIHC indicators have shown applicability and reliability for tracking the consequences of EP (Cheng et al., 2021; Nie et al., 2021; Shi et al., 2022b).

Because WS is a result of the socioeconomic and natural environment, making it an exogenous factor, individual households cannot influence regional WS. Therefore, this study did not consider the endogeneity problem of the effect of WS on EP. A specific set of control variables was selected to consider and adjust for household socio-economic attributes, as indicated in previous studies (Nie et al., 2021; Shi et al., 2022a). The control variables for this study were as follows: marital status, home ownership, household size, education, age, gender, employment, and family income. Recent studies have also pointed to the impact of abnormal temperatures on EP (Feeny et al., 2021; Que et al., 2022). Therefore, abnormal temperatures are included in the control variables. As northern China has taken the lead in implementing coal-to-gas and coal-to-electricity policies in areas near provincial capitals in order to combat air pollution caused by heating in rural areas, the cost of heating has risen, thus potentially exacerbating rural EP in areas near provincial capitals (Xie et al., 2022). Therefore, distance from the provincial capital was also used as a control variable in this study. Table 1 presents the specific definitions of the control variables and descriptive statistics.

Equation (2) depicts the structural model of the core variables. Labor costs and fertilizer costs are the channel variables. Labor costs were used to characterize labor loss effects. Fertilizer costs were used to characterize factor substitution effects. The definitions of labor costs and fertilizer costs, as well as descriptive statistics, are shown in Table 1. The symbol X represents a collection of control variables, as mentioned previously.

and are the direct effects of channel variables on EP. is the direct effect of WS on EP. is the indirect effect of WS on EP through the channel variable labor costs. is the indirect effect of WS on EP through the channel variable fertilizer costs.

The path analysis model was estimated in this study using the MLE method. Utilizing the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), and the comparative fit index (CFI), this study evaluated the goodness of fit of path analysis estimates, with the following criteria: <= 0.08, <= 0.1, and >= 0.9 (Schermelleh-Engel et al., 2003).

The percentage of rural households in EP in the study area ranges from 13.1 to 24.6%, depending on the different measures of EP (Table 1).

This study used EP4 as the primary indicator of EP. The baseline estimates in Table 2 show the pooled probit estimates of WS on EP, with a positive and statistically significant effect of WS on EP. The results remain significant after replacing the different EP indicators. For the control variables, household income negatively affects EP and is statistically significant, and employment status negatively affects EP and is statistically significant. When the household income increases, the household EP significantly decreases. When the head of the household owns the job, the significant degree of EP in the household is significantly reduced. This indicates that an increase in household income and the household being in employment status reduces the EP status of the household. The effect of homeownership on EP was not significant. However, household size significantly increased EP. A possible explanation is that larger household size means that households are more likely to have elderly people, who spend more time indoors and consume more energy. Abnormal temperatures also increase EP in households, and households may increase their energy bills in order to achieve comfortable temperatures. The closer the distance to the provincial capital city, the higher the likelihood of EP. This may be due to higher energy prices caused by the strong push for cleaner rural energy transitions in neighboring provincial capitals, which increases household energy expenditures and thus increases the risk of EP. The other variables do not show the expected direction of impact or are not statistically significant.

Table 3 shows the results of the robustness analysis, which was estimated using a Logit model. With the exception of the EP3 indicator, which happens to fail the 10% significance test level, the results of the estimation of the three EP indicators are that WS positively affects EP at the 1% significance level.

As shown in Figure 7, WS has a significant positive effect on labor costs with a coefficient of 0.0314, which is significant at the 1% level. WS has a significant positive effect on fertilizer costs with a coefficient of 0.0297, which is significant at the 1% level. Labor costs have a significant positive effect on EP, with a coefficient of 0.0588, significant at the 1% level. Fertilizer costs significantly and positively affect EP with a coefficient of 0.0362, which is significant at the 1% level. The direct effect of WS on EP is significant at the 5% level, whose coefficient is 0.0252.

After substituting various EP indicators, Table 4 displays the results. Consistent with the sign of the coefficients, the study yields the same results. This demonstrates the reliability of the study's findings. The conclusion of the mechanism analysis is that WS increases EP through labor loss effects and factor substitution effects. WS has a direct effect on EP.

WS compels smallholder farmers to utilize additional inputs to prevent diminished agricultural yields. Due to the lack of social security for smallholders compared to urban residents, agricultural production is frequently perceived by smallholders as a final means of survival. Consequently, their decisions reflect greater survival rationality than economic rationality. Survival rationality prioritizes the minimization of agricultural production decline over the maximization of economic returns. The rationale for action within the framework of economic rationality appears to involve acknowledging the irrecoverable costs associated with diminished agricultural output resulting from WS and pursuing greater income compensation in alternative non-agricultural sectors. In the unique context of rural China, theoretical economic rationality contradicts the actual circumstances. EP is a consequence of WS, according to the logic of survival rationality. This is a significant observation from the mechanism analysis.

This study identified a novel environmental factor affecting energy: WS, despite the literature review detailing numerous existing factors (Sadath & Acharya, 2017; Winkler, 2017; Liu et al., 2018; Qu & Hao, 2022; Kyprianou et al., 2023; Okyere et al., 2023). We have contended that a detrimental socio-economic effect of WS is the exacerbation of EP in rural regions (Oswald Spring, 2011; Mekonnen & Hoekstra, 2016; De Almeida Castro et al., 2018; Ikuemonisan et al., 2021; Castelvecchi, 2023). This serves as a valuable enhancement and expansion of two bodies of literature: one addressing the determinants of EP and the other examining the ramifications of WS.

Despite the scarcity of relevant studies addressing the same topics as this research, the current study essentially reaches analogous results to the pertinent research. A study in Zimbabwe observed that the effects of water and energy on homes are intricate; nonetheless, alleviating WS enhances the energy conditions (Gandidzanwa & Togo, 2024). Water supply necessitates considerable energy expenditure (Gandidzanwa et al., 2024). A study in Iran indicated that water constraint results in diminished energy efficiency, namely that irrigation requires more energy (Soltani et al., 2023). This reasoning aligns with the study's premise that WS elevates irrigation energy usage, thereby resulting in EP. It has been noted that WS results in an escalation of other agricultural input parameters to alleviate the detrimental impacts of WS (Feng et al., 2022), closely paralleling the findings of the mechanism study component of this research. The study's mechanism is examined from a micro perspective, emphasizing household survival reasoning over economic rationality. The principal findings of this study closely resemble those of analogous research, exhibiting a significant degree of comparable internal reasoning. This work expands upon this basis by examining the effects of WS on EP and by developing a comprehensive mechanistic analysis.

This study integrates satellite data from the GRACE with socioeconomic data from the CFPS to examine the impact of WS on EP among rural households in northern China. By addressing a critical gap in the literature, this study provides valuable insights into how WS exacerbates EP. The main findings are as follows:

• 1. WS in northern China has intensified in recent years.

• 2. The incidence of EP among rural households in the sample remains largely unchanged over time.

• 3. WS has a statistically significant positive impact on EP.

• 4. WS increases agricultural production costs, thereby contributing to EP, particularly through labor loss and factor substitution.

Based on these findings, the study proposes the following policy recommendations:

• 1. Improve irrigation efficiency: Prioritize the development of water-saving infrastructure and energy-efficient irrigation systems in water-scarce rural areas to reduce irrigation costs and mitigate EP.

• 2. Energy consumption subsidies: Increase subsidies for energy consumption in low-income rural households, particularly in areas where WS significantly increases energy costs. The design of these subsidies should be refined through cost–benefit analyses based on local conditions.

• 3. Irrigation electricity subsidies: Implement subsidies for irrigation electricity costs to reduce the financial burden on households in areas heavily dependent on irrigation. These subsidies should be evaluated based on energy consumption patterns and specific regional needs.

• 4. Support agricultural inputs: Provide subsidies for agricultural inputs, such as fertilizers, in regions where WS cannot be addressed in the short term. This will help alleviate the financial burden on farmers who are compensating for reduced yields.

• 5. Future research: Conduct field-based cost–benefit analyses to evaluate the impact of irrigation-related subsidies and assess their effectiveness in mitigating EP. Further research is needed to explore the cost-effectiveness of these policies and their long-term sustainability.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.