A simultaneous equations approach to analyze the sustainable water–energy–food nexus in South Korea

Factors such as climate change, economic development, population growth, pandemics, and geopolitical instability threaten water, energy, and food (WEF) security, which consequently put sustainability at risk. However, studies that simultaneously consider WEF security and sustainability aspects still need improvement. This research aimed to build a sustainable WEF nexus framework and analyze the interrelationships among water consumption, electricity demand, food production, and ecological footprint, considering the Environmental Kuznets curve (EKC) hypothesis and external factors of the WEF nexus. For the empirical analysis, this study employed the three-stage least squares method to identify synergies and trade-offs in the sustainable WEF nexus in South Korea using panel data from 2005 to 2019. The results indicated that rice production causes excessive use of agricultural water, thereby deteriorating water availability and quality. This phenomenon leads to scarce water resources and environmental degradation, which negatively impact energy production and sustainability. Although increased agricultural productivity through automation improves food security, it can pose a threat to energy security by increasing electricity demand and energy imports. The EKC hypothesis test revealed that environmental problems cannot be solved through economic development. However, the indicators related to WEF security influence environmental sustainability rather than economic growth. These results indicate that WEF security and sustainability can be improved simultaneously by maximizing synergies and minimizing trade-offs within a sustainable WEF nexus. Therefore, this research provides a roadmap for policymakers regarding efficient ways to improve environmental quality and WEF security.


Introduction
The demand for water, energy, and food resources that make up the basic needs of humans has increased globally, and this trend is predicted to continue through 2030 (FAO et al 2021).According to the NIC (2012), the demand for water, energy, and food will increase by 35, 40, and 50%, respectively, in 2030 compared to those in 2012, due to a soaring population, urbanization, and an additional three billion middle-class people by 2030 (WWF SABMiller 2014, Ferroukhi et al 2015).Additionally, the economic, social, and environmental impacts of the COVID-19 pandemic (Al-Saidi and Hussein 2021, Hamid and Mir 2021, Mofijur et al 2021), the threat of climate change (Misra 2014, IPCC 2022), and geopolitical instabilities, such as the Russia-Ukraine war (Liadze et al 2022, Shumilova et al 2023), have aggravated these challenges (Estoque 2022).
To ensure stable access to water, energy, and food sources, further research is necessary to guarantee responsible and sustainable resource management (Peña-Torres et al 2022).These three resources are highly interrelated and thus should be consider together (Daher and Mohtar 2015, Albrecht et al 2018, Brears 2018).For example, water is required for drinking, agricultural irrigation, and the food industry (FAO 2017).Water also plays an important role in energy generation processes, such as the production of hydroelectric power and biofuel, cooling of nuclear and geothermal power plants, and extraction of traditional fuels, notably shale gas development and mining (International Energy Agency (IEA) 2020).Pumping water for food and irrigated agriculture, desalination, water purification, water distribution, wastewater treatment, long-distance water pumping, food production, and its supply chain all require considerable amounts of energy (FAO 2011, WWAP 2014).Food production is an energy and water-intensive industry that consumes significant amounts of energy and water (Bazilian et al 2011, Compton et al 2018).Additionally, bio-crops can be used as a renewable bioenergy source in addition to being a food source (IRENA 2019).
The water-energy-food (WEF) nexus has been used to analyze the complex interrelationship between water, energy, and food resources since the 2011 Bonn Conference on Water, Energy, and Food Security Nexus (Hoff 2011).According to the WEF nexus framework, a choice made regarding the management of one of the three resources will affect choices made for the two other resources (Putra et al 2020, Peña-Torres et al 2022).This idea includes the notion that the supply and demand chains for these resources are all closely intertwined (Bizikova et al 2013, Ringler et al 2013, Rasul 2014).Specifically, WEF proposes a method to maximize synergy and minimize trade-offs by analyzing the complex interactions among the three resources (World Economic Forum 2011, Mukuve andFenner 2015, An 2022).A comprehensive evaluation will cover water, energy, and food as well as environmental, social, and economic drivers, which will also allow for the identification of interrelationships (synergies and trade-offs) across sectors to guide solid WEF-related management and development activities (McCarl et al 2017).
In this context, this study aimed to execute an interrelationship analysis of internal/external drivers and direct/indirect impacts on the sustainable WEF nexus in South Korea through a simultaneous equations approach.This aim is divided into three parts: (1) develop a theoretical basis for a simultaneous equation model (SEM) based on a sustainable WEF nexus framework, (2) explore the interrelationship between WEF security and sustainability by analyzing coefficients between factors involved in the sustainable WEF nexus, and (3) investigate sustainability in South Korea by combining the WEF nexus and the Environmental Kuznets Curve (EKC) hypothesis.The main contribution of this study is the assessment of the WEF nexus in South Korea via systematic estimation that considers perspectives for the achievement of sustainable development.In particular, the employment of the SEM enables the determination of the relationships between numerous variables that have an effect on each other directly or indirectly thereby assessing the effect of different policy interventions and evaluating hypotheses on the causal nexus among variables by systematically considering all relevant interdependencies.

Literature review
The nexus approach has drawn attention in the academic, political, and industrial fields due to its characteristics (Ozturk 2015, Garcia and You 2016, Al-Riffai et al 2017, Weitz et al 2017, Fayiah et al 2020).Dynamic quantification techniques are necessary to identify the critical variables influencing the performance of the coupled nexus system and to highlight the dynamics of natural processes in conjunction with various dimensions that sustain the interrelationships between nexus sectors (Zhang et al 2018).Many earlier studies have analyzed the interactions in the WEF nexus using quantitative methods (Newell et al 2019).

Water-energy-food (WEF) nexus
At the national level, the nexus approach has been applied within the scope of a single country or in a comparison between several countries.Howells et al (2013) presented an integrated assessment model named climate, land use, energy, and water strategies (CLEWs) to identify trade-offs, synergies, and co-benefits by assessing the resource system in Mauritius.Mohtar and Daher (2014) developed a scenario-based tool (WEF Nexus Tool 2.0) to analyze interactions in WEF security while considering social, environmental, and economic changes.Al-Ansari et al (2015) considered Qatar and examined the interrelations between the WEF nexus, focusing on a food production system using a life cycle assessment.Moreover, Owen et al (2018) calculated the consumption-based WEF of the UK and explored the critical supply chain by considering the entire lifetime of a product via an input-output analysis.Zhou et al (2016) built a computable general equilibrium model with a tax module to study China's nexus system.Campana et al (2018) investigated the effects of drought on the requirements for WEF resources during a drought period in Sweden by using a multi-objective simulation-optimization model.In another study, Putra et al (2020) confirmed the interactions between the WEF sectors in South Asian countries (Bangladesh, India, Nepal, Pakistan, and Sri Lanka) through correlation and network analyses.Wicaksono and Kang (2019) developed a computer simulation model based on a system dynamics algorithm for the interconnection of WEF resources in South Korea and Indonesia.Huang et al (2020) built China's local WEF nexus through a simultaneous equations approach, involving core, peripheral, and interactive nexuses.These studies revealed that sustainable development can be achieved by exploring the interactions within the WEF nexus.This is because water (Goal 6: clean water and sanitation), energy (Goal 7: affordable and clean energy), and food (Goal 2: zero hunger) resources are the central elements of the Sustainable Development Goals (SDGs), and pressure on these three resources can threaten sustainable development (Bleischwitz et al 2018, Liu et al 2018, Simpson and Jewitt 2019, Akinsete et al 2022).

Environmental Kuznets curve (EKC) hypothesis in relation to WEF nexus
The EKC hypothesis is used to evaluate sustainability (Hartman and Kwon 2005, Farhani et al 2014, Sarkodie and Ozturk 2020).Kuznets (1955) claims an inverted U-shaped relationship between economic development and environmental degradation.The inverted U-shaped curve signifies that environmental conditions deteriorate in the early stages of economic growth and improve in the later stages of economic growth.This implies that environmental degradation initially increases and decreases as the economy grows.Ozturk (2015) conducted a sustainability assessment of the WEF nexus, utilizing dynamic panel modeling with the EKC hypothesis among BRICS countries (Brazil, the Russian Federation, India, China, and South Africa).Moreover, Zaman et al (2017) confirmed the carbon fossil-methane EKC of sub-Saharan African (SSA) countries by analyzing the non-linear relationship between WEF resources and air pollutants.Nassani et al (2019) used the simultaneous generalized method of moments and investigated the relationships between WEF resources, carbon fossil-GHG emissions, and growth-specific factors to verify the EKC in Pakistan.Xu et al (2022) examined the link between economic growth and the WEF footprint by exploring the existence of the EKC in China's economic zones and regions.

Literature gap
Despite the increasing literature on the WEF nexus, the application of nexus frameworks to policy recommendations (Gain et al 2015, Pahl-Wostl et al 2018, Silalertruksa and Gheewala 2018, Olawuyi 2020, Lazaro et al 2022) and the consideration of external drivers (environmental, economic, and social dimensions) in the WEF system are still limited (Räsänen et al 2015).Systematic studies that explain the interaction of WEF security, considering indirect impacts, are lacking (Wicaksono and Kang 2019).Furthermore, there is a lack of research on sustainability, although the WEF nexus is closely related to sustainable development.Finally, most studies on the EKC hypothesis rely on atmospheric indicators, whereas the literature on the EKC hypothesis that uses land, biodiversity, and freshwater indicators is erratic and sparse (Sarkodie and Strezov 2019).
Therefore, applying the SEM and considering sustainability will help in the establishment of policies for the sustainable WEF nexus by simultaneously analyzing bi-directional interactions caused by external drivers and indirect impacts (Ozturk 2016, Galdeano-Gómez et al 2017, Fan et al 2018).

Research area
This study was conducted at the national level in South Korea, which has considerable economic and population growth with limited resources.South Korea, formally the Republic of Korea (ROK), is located in Northeast Asia, with a total land area of 97.6 × 10 3 km 2 .In 2022, the total population of South Korea was 52 million and the gross domestic product (GDP) was 1.67 trillion USD.The country's dependence on energy resource imports is 95% (Korea Energy Economics Institute 2019), and its food self-sufficiency rate is only 23% (Korea Rural Economic Institute 2019).Moreover, it has a considerably high water footprint (Hoekstra and Chapagain 2011), which, together with the deterioration of water infrastructure (Kang 2019), threatens water security.Therefore, the availability aspect of WEF resources is particularly vulnerable (Simpson et al 2022).Integrating the interaction between WEF security and the nexus approach can prove helpful in achieving sustainable resource management in society.

Frameworks and variables
Various frameworks have been suggested according to the purpose and scope of the study, but the following are the cornerstones of the WEF nexus framework.Hoff (2011) presented a WEF nexus framework centered on water supply, energy, and food security, all of which are connected to water availability.It also considers global trends, including urbanization, population, and climate change, to promote WEF security, sustainable growth, and a productive environment.The World Economic Forum (2011) offers a framework in which water and food security are connected to economic disparity and global governance failures, as well as energy security, causing chronic WEF shortages and crises.This comprehensive structure includes external drivers affecting the nexus, such as demographic, economic, and environmental factors.The Food and Agriculture Organization (FAO)'s WEF framework addresses the interrelations between human and natural systems, focusing on biophysical and socioeconomic resources related to the WEF nexus (Flammini et al 2014).These interactions are affected by external global drivers, such as population change, urbanization, climate change, industrial development, and sectoral policies.
Based on the existing literature on the WEF, the author built a systematic framework for a sustainable WEF nexus by coupling it with sustainability at the national level (figure 1).In this framework, WEF security and sustainability, which are core and internal factors, are not only influenced by each other's security but also by external factors, such as the environment, society, and economy.The criteria for the WEF security indicators considered availability, accessibility, affordability, and productivity drivers, following the study by An (2022).Each indicator can meet more than one criterion and affect other internal nexus factors.For example, agricultural productivity (AP) not only affects the accessibility, affordability, and productivity of food security, but also the availability of water and energy security.This influence also has a ripple effect on sustainability, which is linked to WEF security.Ecological footprint was selected as an indicator of sustainability.It identifies the use of productive surface areas, which consist of cropland, grazing land, fishing grounds, built-up land, forest areas, and carbon demand, and assesses how much is needed to produce various resources that humankind consumes and disposes of as waste (Wackernagel and Rees 1998, Moffatt 2000, Wackernagel et al 2021).
To examine the interrelationships in the proposed framework, the author collected a panel dataset for 2005-2019 (table 1).Appendix presents the descriptive statistics of all variables for the entire period.Data were obtained from the IEA, UN, Organization for Economic Cooperation and Development (OECD), Global Footprint Network (GFN), Ministry of Environment of South Korea (MOE), Korean Statistical Information Service (KOSIS), Ministry of Agriculture, Food and Rural Affairs of South Korea (MAFRA), and the National Groundwater Information Management and Service Center (GIMS).

Model specification
The SEM denotes a system of linear equations that includes a feedback relationship between variables, where some variables occur as explained in one equation, whereas others may appear as explanatory variables (Wooldridge 2012).The variables of an SEM may be connected through direct relationships, indirect ties, reciprocal interactions, feedback loops, and correlations between disturbances (Maddala 1992, Paxton et al 2011).The two most popular estimation methods for SEM are two-stage least squares (2SLS) and three-stage least squares (3SLS) (Kapteyn and Fiebig 1981).The 2SLS, which is a single-equation approach, implies that overidentifying restrictions in other equations is not considered when estimating the parameters in a particular equation.In contrast, 3SLS, which is a system equation approach, uses information concerning the endogenous variables in the system and considers error covariances across equations, hence, it is asymptotically efficient in the absence of specification errors (Greene 2008).A systemic method can also help consider important indicators, such as sustainability and resilience, with regard to linkages across different domains rather than just as their respective components (Huntington et al 2021).The endogeneity issue in the SEM estimation can also be resolved by using the 2SLS approach.However, if the variance-covariance matrix of the disturbance is not diagonal, the estimators from the 2SLS cannot be asymptotically effective (Zellner and Theil 1992).As a combination of 2SLS and seemingly unrelated regression (SUR), 3SLS will guide consistent and efficient estimates if the disturbances are correlated contemporaneously (Henningsen and Hamann 2008) because SUR may enhance the efficiency of parameter estimates in the presence of contemporaneous correlation of errors across equations.In addition, the 3SLS is often, but not always, superior to the 2SLS, especially in over-identified equations (Kennedy 2008, Larcker andRusticus 2010).Therefore, in this study, 3SLS was used to examine the direct and indirect relationships in the WEF nexus framework.
In this study, four equations for interrelationships in the internal and external nexus based on the abovementioned literature, structural concept, and collected data in section 2.2 were constructed.Water consumption, energy demand, food production, and ecological footprint were classified as endogenous variables, whereas other variables were categorized as exogenous.The four endogenous variables are indicators that play key roles in each sustainable WEF nexus (Hoff 2011, Bizikova et al 2013).Moreover, the security of one where t indicates the time span from to 2005-2019 and u denotes the error term.Equation (4) confirms the EKC by analyzing the relationship between environmental sustainability and economic growth.Including economic growth in equation (4) helps explore the EKC hypothesis.
The system of equations should be over-identified in most simultaneous equation models (Zellner and Theil 1992).Every equation of the system obeys the rank and order conditions for identifiability (identified and overidentified).Moreover, to understand the analysis results of 3SLS systematically, all variables were transformed into logarithmic scales (equations ( 1)-( 4)).The estimated coefficient is the elasticity obtained by adopting the logarithmic form (Auster et al 1972).For example, if an independent variable exhibits a negative coefficient (b), a 1% increase in the value of the independent variable decreases the value of the dependent variable by b%.Using this mathematical structure, the interrelationship of the WEF nexus is evaluated in section 3.

Model verifications
Prior to interpreting the 3SLS results, several tests were conducted to verify the appropriateness of the proposed model.First, the Wu-Hausman test was used to detect endogenous regressors for the suggested four equations (Hausman 1978).Second, the validity of the overidentifying restrictions was verified by applying the Sargan-Hansen test (Sargan 1958).Third, the Wald test, which is a parametric statistical measure, was conducted to confirm whether the independent variables are significant for the model (Wald 1943).Finally, the variance inflation factor was used to detect multicollinearity (Alin 2010).As shown in tables 2 and 3, the results of the four tests indicate that the model is appropriate.

Empirical findings
Table 4 shows the estimation results for the system of simultaneous equations with water consumption, energy demand, food production, and ecological footprint as endogenous variables.The goodness of fit, measured by the overall R-squared value of 0.9592 in all equations, shows that the explanatory variables are sufficient to explain changes in the endogenous variables across South Korea.Figures 2 and 3 were visualized using only the statistically significant values from the results in table 4. The interrelationships for each indicator consist of three parts: positive links representing synergies, negative links indicating trade-offs, and intrinsic links representing inherent connections.Intrinsic links are used according to the theoretical background because the interrelationships between the indicators are difficult to explain.This relationship can be interpreted using the principle of the nexus approach.For example, the consumer price index for energy (CPIE) cannot explain the direct effect on electricity demand (ED), however, it indirectly affects ED because it affects AP.    -water 2023).Therefore, agriculture, especially rice production, accounts for a significant portion of the country's water resources (Yoo et al 2014).The introduction of technologies and policies that increase agricultural water use efficiency (Wallace 2000, Howell 2001, Hsiao et al 2007) and the production of crops with a lower water footprint than rice (e.g., potato, sweet potato, taro) (Mekonnen and Hoekstra 2011) will improve water and food security simultaneously (Davis et al 2017).
The climate disaster cost (CD), which escalates due to climate change (Banholzer et al 2014), exerts a significantly positive effect on ED (0.0475).This relationship can be explained by the fact that increasing global average temperatures are associated with widespread changes in weather patterns (IPCC 2022).Extreme temperature fluctuations increase the demand for heating and cooling, leading to an increase in electricity demand (Parkpoom et al 2004, Franco and Sanstad 2008, Eskeland and Mideksa 2010, Allen et al 2016, Auffhammer et al 2017).
AP was significant and positive for ED (2.7646) and food production (FD, 0.5373).Agricultural automation, which has been recently introduced, helps improve food productivity by reducing labor costs, improving crop productivity, and decreasing working hours (Edan et al 2009, Kim et al 2020, O'Shaughnessy et al 2021).Hence, the share of electricity in agriculture increased from 11.7% in 2001 to 40% in 2019, whereas that of oil decreased from 85.9% to 57.3% over the same period (Korea Energy Economics Institute 2022).The rate of increase in the use of agricultural electricity has been 7.4% per year on average since 2005, far exceeding the rate of increase in total electricity consumption (5.7%), this trend is expected to increase further in the predictable future (KOSIS 2022).As Korea heavily relies on imports of energy resources for power generation, it also relies on energy imports for food security, even though it increases domestic food production in the current situation.This phenomenon explains why CPIE has an intrinsic connection to AP.The use of renewable energy, a sustainable power supply, in agricultural areas can maximize agricultural productivity and minimize electricity demand, thereby overcoming the abovementioned problems (Ravi et al 2016, Aghajanzadeh and Therkelsen 2019, Gorjian et al 2020, Gorjian et al 2022).
The ecological footprint (EF) increased by 31.309% when rice production increased by 1%.In Korea's food consumption pattern, per capita rice consumption continues to decrease, and meat consumption increases (KOSIS 2022).The government has provided huge subsidies for rice production through a direct payment system to support the supply of rice, which is a staple food (OECD 2022b).As a result, farmers have flocked to rice production because it guarantees stable income, resulting in an oversupply of rice since the 2000s.In contrast, the self-sufficiency rates for crops other than rice, such as wheat (1.2%), corn (3.3%), beans (25.4%), and barley (32.6%), are very low (MAFRA 2021).The oversupply of food worsens food security and environmental conditions by causing food waste (Messner et al 2021).In Korea, where only an oversupply of rice is observed, this effect is more severe.According to Xu et al (2021), rice production in paddy fields generates the highest carbon emissions among all plant-based foods.If subsidies for rice oversupply can be invested in crop diversification, food security can be improved while minimizing environmental damage (Smith et al 2008, Massawe et al 2016, Renard and Tilman 2019).
A 1% increase in both low-carbon power (LCP) and the public energy RD&D budget (PEB) has a positive impact on EF by 21.821 and 23.603%, respectively.This result is inconsistent with the findings of Garrone and Grilli (2010), Brouwer et al (2016), Anadón et al (2017), Pehl et al (2017), Zeyringer et al (2018), andZhu et al (2021), who found that a low-carbon power system and public energy RD&D investment promote a decarbonized society and contribute to sustainability by improving energy efficiency and reducing carbon intensity.Over the past decade, the Korean government has heavily invested in renewable energy, particularly solar power (IEA 2020).In addition, investments in low-carbon power and public energy RD&D have focused on solar photovoltaics.As a result, the renewable energy mix in 2021 represented solar, hydroelectric, wind, and wave and tidal powers at 69.5, 19.8, 9.3, and 1.3%, respectively, indicating that solar power will account for most of the renewable energy generation (IEA 2022a, IEA 2022b).However, solar power may lead to land occupation and change, which can adversely affect ecosystems and biodiversity (Turney andFthenakis 2011, Wu et al 2021).Korea does not consider the aforementioned aspects of deploying solar power.The indiscriminate installation of solar power can damage farmland and forests, consequently destroying the ecosystem and offsetting the benefits of reducing carbon.Therefore, solar energy that minimizes environmental burdens, such as building-integrated photovoltaics (BIPV) (Peng et

Trade-off context
Figure 3 shows the estimation results with trade-off relationships, except for the statistically insignificant values.Energy efficiency RD&D budget (EEB) indicates a negative relation with WC (−0.0257).An increase in energy efficiency indicates a decrease in energy input.Approximately 90% of South Korea's energy mix in 2021 will be non-renewable energy, such as nuclear, LNG, and coal (KOSIS 2022).These non-renewable energy sources use significant amounts of water in power generation, particularly during cooling (Spang et al 2014, Qin et al 2015, Lee et al 2018, Larsen and Drews 2019).In contrast, renewable energy (solar and wind) consumes a small amount of water for cleaning, whereas non-renewable energy uses more than 500 gal MW −1 h −1 of water for cooling (Macknick et al 2012).
Economic growth (EG) is significantly negative for FP (−0.0404).As the economy grows, meat consumption also tends to increase (Gerbens-Leenes et al 2010, Sans and Combris 2015).South Korea has continued economic growth, and meat consumption per capita has also increased by more than 20 kg, from 31.9 kg in 2000 to 53.9 kg in 2018 (MAFRA 2021).Although meat productivity has improved, it has not sustained the rising meat consumption and has begun to rely on imports.The meat self-sufficiency rate decreased by 14.6% from 78.8% in 2018 to 64.2% in 2000 (MAFRA 2021).Consequently, as the economy grows, it becomes dependent on food imports, and domestic food production decreases.CI exerts a significant negative effect on FP (−0.9609).Agricultural water accounts for 62.3% of total water consumption (MOE 2022).Limited water availability and increased water consumption for non-renewable energy due to the abovementioned positive relationship between CI and WC will negatively impact agricultural water use.This finding is consistent with that of Wicaksono and Kang (2019), who found that securing water availability by decreasing the share of nonrenewable energy will help food security in the future.As the EF value increased and became a threat to sustainability, it had a negative impact on FP (−0.0093).Given that it entails the use of land, water, and energy, and the creation of trash, food production is one of the key contributors to the ecological footprint (van Noordwijk and Brussaard 2014, Galli et al 2017).Hence, an increasing ecological footprint implies that more resources are being used and more waste is being generated, which can deteriorate the quality and availability of water and land for food production (Wackernagel andRees 1998, Wackernagel et al 2021).A 1% increase in lake and river permanent water areas (LRA) will directly result in a 1.49 and 52.65% decrease in FP and EF, respectively.Agriculture is a water-intensive industry closely related to water availability (Postel 1998, Wallace 2000, Gordon et al 2010).Globally, approximately 70% of surface water consists of lakes and rivers (Shiklomanov and Rodda 2003).An increase in the area of lakes and rivers can negatively impact food production if the cost of pollution exceeds water availability.In South Korea, nonpoint pollution sources pollute water resources owing to urbanization and a steady increase in meat production, this trend is gradually increasing (OECD 2018a).In contrast, the ecological footprint can be reduced by increasing the area of lakes and rivers because freshwater environments clean and store water, which is crucial for people and ecosystems (Kitzes et al 2007, Kitzes andWackernagel 2009).Consequently, the simultaneous management of water quality and quantity is necessary for sustainable food production and ecosystems (Kirby et al 2003).
The negative relationship between population growth (PG) and EF does not correspond to the findings of Dietz et al (2007) or Kitzes et al (2008), who suggest that PG consumes more resources and adds to the environmental burden.Generally, as the population decreases, the number of resources used also decreases.However, if unsustainable consumption continues, such as overconsumption of water, over emission of carbon, use of fossil fuels, and a diet centered on meat consumption, the environmental burden will be greater than the declining resource consumption (Spangenberg and Lorek 2002, Jackson and Papathanasopoulou 2008, Hoekstra and Wiedmann 2014).Groundwater for agriculture (GWA) has a trade-off relationship with EF.Groundwater is a crucial source of water for irrigation, and accounts for more than 40% in many OECD nations and 70% of water use globally (Gruère and Shigemitsu 2021).In 2020, South Korea's agricultural sector accounted for approximately 53% of total groundwater use (GIMS 2022).In the agriculture sector, groundwater use is more efficient in terms of energy consumption than using surface water owing to more streamlined distribution, transportation, and supply processes (Siddiqi andAnadon 2011, Yang et al 2016).This can reduce the carbon footprint and thus help diminish the ecological footprint in the short term.However, land subsidence, saltwater intrusion, and other environmental issues can result from intensive groundwater pumping for agriculture, which depletes aquifers in the long term (Hallberg 1986, Han 2003, Raquel et al 2007).Groundwater is considered a finite resource because aquifer recharge rates are frequently sluggish (Merchant 1994, Madramootoo 2012, Richey et al 2015).Hence, controlling the use of GWA while considering the environmental burden and availability caused by the consumption of subsurface water is important (Shan et al 2009).
A 1% increase in low-carbon energy technology (LE) decreased ED by 0.3563%.High electricity prices have suppressed electricity demand growth worldwide (IEA 2023).South Korea's levelized cost of electricity (LCOE) is as follows (Lorenczik et al 2020): gas, 90.19-100.43 USD/MWh, coal, 81.04 USD/MWh, nuclear, 67.16 USD/ MWh, wind (offshore), 119.31USD/MWh, solar (commercial), 121.14 USD/MWh, wind (onshore), 137.02USD/MWh, and wind (offshore), 193.24USD/MWh.Owing to continued investment in clean energy, the LCOE of renewable energy is declining faster than that of non-renewable energy.For example, solar power in South Korea is expected to achieve grid parity by 2025 (Hong et al 2020).Hence, the negative relationship between LE and ED is expected to increase gradually.Consumer price index for food (CPIF) is intrinsically linked to EG and negatively affects ED.Maintaining stable food prices is an important factor in the economic growth of Asian countries (Dawe and Timmer 2012).This is because the ripple effect of economic and social costs resulting from price instability slows economic growth (Byerlee et al 2006, Jayne 2012, Verpoorten et al 2013).For example, when food prices rise, households reduce their consumption of goods, which worsens the economic cycle.Contrary to previous studies that showed that food production and electricity demand have a positive relationship (Khan andHanjra 2009, Ladha-Sabur et al 2019), ED decreased by −0.7029% when FP increased by 1%.This is because energy consumption in the agricultural sector consists of 57.3% oil and 40% electricity, and agricultural productivity is continuously increasing (Korea Energy Economics Institute 2019).Additionally, power savings in the food industry, which are directly related to food production, have steadily increased since 2013 (KOSIS 2022).However, the potential for productivity decline due to the aging rural population and the increasing share of electrification in agricultural energy use remain issues to be addressed (OECD 2018b).WC and industrial water consumption have intrinsic connections to LRA and GWA, respectively.When water is removed from rivers and lakes for drinking, the water flow may decrease, which reduces the total area of the water body (Wurtsbaugh et al 2017).The increased evaporation rate owing to climate change will further accelerate this situation (Shenbin et al 2006, Woolway et al 2020).Water availability in a country is limited by factors, such as precipitation and water resource management (Chenoweth 2008, Elliott et al 2014).Increased water use in the industrial sector reduces the availability of water resources for municipal and agricultural use.

EKC hypothesis
As described in table 4, the empirical findings of this study cannot establish the EKC hypothesis, as is evident from the EG estimates, which depict a statistically insignificant relationship with EF.This conflicts with the findings of Iwata et al (2012), Onafowora and Owoye (2014), Onater-Isberk (2016), and Destek and Sarkodie (2019), who found that the EKC hypothesis is valid for South Korea.The rationale behind this discrepancy is that the present study considered indicators associated with WEF security that were neglected by previous studies.Rejecting the EKC hypothesis implies that economic growth does not guarantee environmental sustainability.Thus, other actions are required to address critical environmental problems (Ali et al 2017, Gill et al 2018, Pata et al 2022).Moreover, focusing on the scale-up of economic factors and other external factors will undermine the improvement of WEF security (Huang et al 2023).According to the aforementioned discussion (Sections 3.2.1 and 3.2.2), the components of WEF security (LCP, RP, PEB, LRA, and GWA) impact environmental sustainability rather than economic development.Maximizing synergies and minimizing trade-offs among WEF resources improve WEF security and achieve environmental sustainability.This finding is consistent with those of Zaman et al (2017), Zaman (2018), Nassani et al (2019), andXu et al (2022), who found that environmental quality can be enhanced through the WEF nexus framework.

Conclusion
In the present study, the sustainable WEF nexus framework was introduced to explore the interrelationships between WEF security and sustainability under a simultaneous equation model and the EKC hypothesis in South Korea during 2005-2019.The key findings and relevant policy implications of this study are as follows.First, rice production is a water-and energy-intensive industry with low production per unit area that adversely affects the sustainable WEF nexus.Excessive use of agricultural water deteriorates water availability and quality, resulting in scarce water resources.Moreover, environmental degradation negatively affects energy production and sustainability.Korea's rice-oriented food production structure accelerates this effect; therefore, expanding the production of alternative crops, such as potatoes and sweet potatoes, is important.Second, an increase in agricultural productivity caused by automation can improve food security; however, it can also pose a threat to energy security by increasing electricity demand and energy imports.The share of renewable energy sources must increase to achieve stable food production.Compared with non-renewable energy, renewable energy consumes little water and does not emit pollutants; therefore, it can positively impact the sustainable WEF nexus.However, South Korea's renewable energy industry focuses on solar power, and the current solar power policy is harming sustainability because installations can damage nature.Solar power sources, such as agrivoltaic farming, float solar power, and BIPV, which do not adversely affect sustainability, should be actively introduced.Third, according to the EKC hypothesis, environmental problems cannot be resolved through economic development.In particular, in the case of South Korea, even if the population decreased because of unsustainable consumption patterns, sustainability was undermined.Policy established on a "wait and grow" presumption is not appropriate (Agras and Chapman 1999), and the current generation should strive for sustainable development.Sustainable WEF security can be achieved by analyzing the synergies and trade-offs of WEF security, a key element of the SDGs, through the Nexus approach.
This study has three limitations.First, a few indicators associated with the sustainable WEF nexus were selected based on data availability and the statistical characteristics of the SEM.Although this study considered more indicators than previous studies, the selected indicators do not thoroughly represent the sustainable WEF nexus.Introducing dimension reduction methods, such as the principal component analysis, allowed the author to handle larger amounts of data.Second, only GDP growth rate was selected as a factor to analyze the EKC hypothesis.Future studies should consider various economic factors, such as GDP, squared GDP, and GDP per capita, for a more in-depth analysis.Lastly, the analysis performed in this study and the suggested opinions are based on historical data.In particular, there is speculation on the development of renewable energy, even though reasonable grounds for the speculation are presented on the basis of existing literature.To handle a broader range of possible results, a procedure for the verification of the effectiveness of proposed measures should be developed in further research.
In conclusion, this study presents a sustainable WEF nexus framework that considers sustainability and external factors in the existing WEF nexus theory.Through SEM and the EKC hypothesis, it was determined that the optimization of synergies and trade-offs between the interconnected sustainable WEF nexus can contribute to sustainable development in South Korea.

Figure 1 .
Figure 1.Framework of the WEF nexus at the national scale.

Figure 2 .
Figure 2. Positive feedback loops of sustainable WEF nexus.

Figure 3 .
Figure 3. Negative feedback loops of sustainable WEF nexus.
al 2011, Shukla et al 2017), agrivoltaic farming (Adeh et al 2019, Miskin et al 2019), and floating solar technology (Oliveira-Pinto and Stokkermans 2020, Hooper et al 2021), should be actively introduced to overcome the paradoxical situation in which solar power damages sustainability.

Table 1 .
(Flammini et al 2014)Artioli et al 2017, Albrecht et al 2018, Endo et al 2020) resources because WEF security interacts with each other(Ringler et al 2013, Artioli et al 2017, Albrecht et al 2018, Endo et al 2020).For example, the FAO has analyzed these indicators to be considered for sustainable water-energy, water-food, and energy-food linkages(Flammini et al 2014).External factors affecting WEF security and sustainability include environmental, social, and economic factors.Considering these relationships and the multicollinearity among the independent variables, the author set up equations as follows:
(Rahman et al 2022)xtFigure2summarizes the estimation results on synergistic relationships, excluding the values that were not statistically significant.Carbon intensity (CI) indicated a positive relationship with water consumption (WC, 0.1347).A high carbon intensity of power generation implies that the proportion of fossil energy sources, such as coal, oil, and natural gas, is high in the energy mix(Rahman et al 2022).These fossil-fuel-based power plants use substantial amounts of water resources for power generation (Qin et al 2015, Stokes-Draut et al 2017, Lee et al 2018), accounting for approximately 60% of South Korea's energy mix.Rice production and agricultural water consumption (AWC) increased WC by 0.2465 and 0.4892%, respectively.Although the food self-sufficiency rate was only 23% in 2018, the rice self-sufficiency rate was 97%.Approximately 80% of agricultural water, which accounts for 61% of the total water use, is used in paddy fields, and rice is grown on most paddy farms (MOE and K