Domestic and International Drivers of the Demand for Water Resources in the Context of Water Scarcity: A Cross-Country Study

Global warming, while increasing human demand for water, is reducing water availability by reducing runoff flows and the effective amount of water between seasons, making water scarcity a growing problem globally. Water management plays an important role in mitigating global warming, improving the water cycle, reducing carbon emissions, and providing clean energy, and pricing water is considered a good approach to water management. Pricing water needs to take into account all sectors and aspects of society, such as domestic water, food and agriculture, energy, transport, industry, urban provision, human health, ecosystems, and the environment, and their interrelationships through water, within the context of the fundamental human rights to water and sanitation. This requires that every stakeholder should contribute to the development of water-related policies at every stage of the water interrelationship. This study investigated the relationship between water demand across different sectors of the economy using indicators for China, Australia, Japan, and the UK. Using panel analyses, this study finds that economic growth and population expansion increases the demand for water in all aspects. These findings have significant policy implications for water management. Because water prices can have an impact on global trade and, more importantly, are a major solution to global warming, water management policies should be considered at the global level, not only at the national level.


Introduction
Global warming 1 affects climate change, ecosystems, living conditions, urban development, and economic development in a variety of ways, and water has become one of the main channels by which human feel these changes. These effects are becoming increasingly evident; for example, between 1901 and 2010, the global mean sea level rose by 0.19 m at the highest rate ever recorded (Church et al. 2013). Even a small amount of mean sea level rise will significantly increase the frequency and intensity of global flooding (IPCC 2019), with about 1.3% of the global population exposed to a 100-year flood (Muis et al. 2016). Increased water temperatures also alter the biogeochemical balance of freshwater ecosystems, leading to water quality degradation through frequent algal outbreaks 1 Global warming is caused by human emissions of large amounts of carbon dioxide and other greenhouse gases that are accumulating in the atmosphere through the burning of fossil fuels, primarily coal, oil, and natural gas (IPCC 2019), resulting in an additional increase in global surface temperatures compared to pre-industrial revolution levels.
Discussion about the efficient allocation of resources has taken place over a long period of time (Koopmans 1951). An efficient allocation of resources is essential to reaching a desirable situation in which our scarce resources are used to produce particular types of goods and services that best maximize the overall satisfaction of society's needs and wants, wellbeing, or living standards. Common access goods, including environmental and natural resources such as air, minerals, oil, forests, river water, and ocean-going fish, are both non-excludable and rival each other. With common access goods, the market fails to send the proper price signals that lead to an efficient allocation of resources. This is a serious problem for society because the survival of current and future generations may be jeopardized when these goods are depleted.
Water, one of the fundamental natural resources, is crucial for social development. Increasing world population, improving living standards, changing consumption patterns, and expanding agricultural irrigation have increased global demand for water (FAO 2013;Mekonnen and Hoekstra 2016;Veldkamp et al. 2015;Kölbel et al. 2018), leading to growing problems of fresh water scarcity throughout the world. Mekonnen and Hoekstra (2016) document that four billion people, i.e., two-thirds of the global population, experience severe water scarcity. Given this water scarcity, in coming decades, water may become the most strategic resource, especially in arid and semi-arid regions of the world.
In response to growing water scarcity, efficient allocation of water resources and proper water pricing become crucial. The key discussion is centered on water rights and water pricing; for example, one group of scholars recognizes water as an economic good (Bauer 2010;Hanemann 2006;Rogers et al. 1998;Rogers et al. 2005). Such shifts have prompted interest in market mechanisms that allocate water for direct human uses and ecosystem needs, involving transfers of established water rights between willing buyers and sellers, based on agreed prices.
Water pricing can be used as an efficient mechanism to manage water uses. Switching to a more appropriate pricing scheme can adjust inefficient levels of water use by changing water demand . For example, Kanakoudis (2002) present residential water conservation techniques that validate the practicality and effectiveness of the operational measures included in a comprehensive residential water conservation program. This urban water pricing policy program was implemented in a rapidly growing area under water-scarce conditions in Athens, Greece. However, implementation of water pricing has been a challenge for both governmental and nongovernmental decision makers. Developing countries, for example, usually suffer from inadequate water supply facilities and lack sophisticated and comprehensive water pricing systems. Thus, they are in need of more practical and effective water pricing methods. Drought and climate change are unevenly distributed geographically, and risks are generally higher for vulnerable populations and communities in countries at all levels of development (IPCC 2014). Moreover, developing countries have less capacity to cope with the impacts of climate change, and the poorest groups and societies are most vulnerable to both small and large shocks. Many developing countries lack financial resources for adaptation and mitigation efforts, and, for some, the ability to act may also be hampered by poor governance (Das Gupta 2013).
Current water pricing mechanisms are in line with the classical economic theory of the supply and demand relationship. The pricing mechanism can be further divided into the supply-side strategy and the demand-side strategy, which focus on balancing the investment and revenue of the water supply service, in addition to capturing the value of water use to users at given prices . Various reasons cause this price inefficiency, including an incorrect estimation of the average price, a low evaluation of the social value of water, a weak enforcement of economic regulation by local governments, and a non-discriminatory price (Garcia and Reynaud 2004).
Because of the growing number of areas facing water scarcity, it is essential to adapt water management strategies beyond traditional water supply and demand management methods . Consumption of water primarily relates to the usage of freshwater and the discharge of wastewater that contains a certain amount of pollutants. Water management can either be done by directly controlling the use of water or by controlling the pollution that can be added to water. Controlling water pollution is similar to controlling environmental pollution via the use of carbon levels, but most attention from scientists and politicians has been paid to local water use efficiency due to the lack of virtual water trade data (Hoekstra and Hung 2002).

Water Consumption in China: An Overview
The pace of economic change in China has been extremely rapid since the economic reforms in 1978. China has accomplished a remarkable feat in transforming itself from one of the poorest countries to becoming the world's second largest economy over the past three decades (Liu et al. 2013). As previously noted, China has an enormous use of resources but water, as one of its most exhaustible resources, has been priced cheaply and used extensively.
Regarding the water scarcity problem in China, the key characteristic is the mismatch between the spatial distributions of water resources, economic development, and other primary factors of production, which leads to the separation of production and consumption of water-intensive products. For instance, the northern, northeastern, and northwestern parts of China account for 57% of land area, but only 15% of the freshwater resource, whereas the eastern, central, and southern areas, with 43% of freshwater resources, account for 57% of China's population and contribute 64% of its GDP (Zhang and Anadon 2014). Furthermore, half of the wheat output in China is produced in the North China Plain, which is heavily dependent on groundwater irrigation in winter. The resulting over-exploitation of groundwater for irrigation has become a major challenge to sustainable social economic development in the arid northern region, having severe adverse impacts on the environment and ecosystems due to drying up of rivers, land subsidence, and sea water intrusion (Foster et al. 2004;Zheng et al. 2010).
The production and consumption of water intensive products are to a large extent separated, and water intensive products are not always produced in regions with water abundance. This study of global water scarcity links the demand for water with its major drivers. Currently, water is not efficiently priced and, in most cases, water prices are kept low for two reasons: water for sustenance and water for political reasons, because higher water prices may be detrimental to the short-term political interests of the policy makers.
Prices, effective in reallocating resources, have also been found to be effective in the event of natural disaster recovery (Rose 2004). China is developing its economy: water prices are relatively low in China and Chinese exports may have benefited from these lower water prices. As such, this study hypothesizes that water consumption has a positive relationship with exports and economic growth in China. Total water consumption in China is related to the economic level of the underlying areas. The consumption of water in China varies both by sector and by region. East China is the most developed of the seven regions and therefore has the highest water consumption, both in terms of total water use and sectors, including agriculture, industry, and households ( Figure 1). Compared to other regions, East China has also performed well in terms of GDP, exports, imports, and the flow of foreign direct investment (FDI) (Figure 2). Analysis of the demand for water is therefore carried out separately for each sector and region. This study also examines the regional distribution in China to determine if the results are different for economically disadvantaged regions, e.g., South East China is economically more advanced compared to the North East and North West regions. The study also accounts for water consumption via foreign investment, in addition to export and import companies. FDI inflows and investment in trading companies are expected to encourage advanced technological development and more efficient use of economic resources including water. J. Risk Financial Manag. 2020, 13, x FOR PEER REVIEW 9 of 27 (a) Percentage of total water consumption by region in China.
(b) Agricultural water consumption as a percentage of total agricultural water consumption by region in China.
(c) Industrial water consumption by region in China as a percentage of total industrial water consumption.
(d) Household water consumption as a percentage of total household water consumption by region in China.

Nature and Measurement of Data
Annual data from 2004 to 2015 (12 observations for each province) was collected from the National Bureau of Statistics of China website database for the major 31 municipalities and provinces within China (Table 1). For empirical analysis, this study divides the Chinese 31 municipalities and provinces into 7 major commonly known regions. The variables included in the model are measured as follows (Table 2): Total water consumption (WCT), total water consumption for agriculture (WCOA), total water consumption for industry (WCOI), and total water consumption for living (WCOL). All of these variables are measured in per hundred million cubic meters. GDP per capita (GDPPC) is GDP hundred RMB per person for each region; population (POP) represents the total population for different regions per 100,000 people; exports (EXP) and imports (IMP) are measured as 100 million USD; foreign direct investment (FDI) is foreign company investment in different regions of China as 100 million USD; sulfur dioxide emissions (SO2) are measured in 10,000 tons. For the purpose of comparison, this study also analyzed the demand for water for a set of developed countries, namely, Australia 5 , Japan 6 , and the United Kingdom (UK) 7 . Annual data for the variables used in the model from 2005 to 2013 (9 observations for each country), were collected from each country's national statistical department website database 8 .This implies that the variables of this study are measured in different units; therefore, before commencing the empirical analysis, it is important to normalize and transfer the data series into natural logarithms to avoid the problems associated with distributional properties of the data series (Paramati et al. 2017

Correlation among Variables
The dependent variable, water consumption, includes WCT, WCOA, WCOI, and WCOL. The independent variables include EXP, GDPPC, IMP, SO2, FDI, and POP. The independent variables with the same nature will be very highly correlated with each other. Therefore, this study derived the correlation matrix for all of the variables to assess the possible problem of multi-collinearity (Table 3). Variables with high correlations were not included in the same model. This study used a number of alternative specifications to avoid the problem of multi-collinearity.

Model Specification
The analysis of the demand for water in China was carried out at the aggregate level for the economy as a whole, and at the disaggregated level. The analysis at the disaggregated level examines the demand for water in the agricultural, industrial, and household sectors. Both agriculture and industry consume water for production purposes, whereas households consume water for the purpose of living. The models used to investigate the drivers of the demand for water in China are specified as follows: Equations (1) to (3) were estimated for China using annual data from 2004 to 2015 9 . Equations (1), (2), (4) and (5) were estimated for Australia, Japan, and the UK using annual data from 2005 to 2013. Water consumption (WC) in Equations (1)-(5) was measured alternatively in terms of WCT, WCOA, WCOI, and WCOL for China, and WCT and WCOL for Australia, Japan, and the UK. The expected signs of the coefficients of different independent variables are given in Table 4. GDPPC is expected to have a positive correlation with water consumption in the early stages of development, when water saving technologies and water management strategies are not very well developed. As an economy develops, it is able to afford better water saving technologies and to have efficient water management strategies. GDPPC is therefore expected to have a negative correlation with water consumption in the early stages of development. The expected sign of the coefficient, therefore, could be positive or negative.

Estimates for China, by Sector and by Region
The empirical analysis began with the panel unit root tests, which were performed to test the stationarity of the variables. To examine the distributional properties and the order of integration of the variables, two generations of tests were undertaken: a first generation IPS (Im et al. 2003) panel unit root test that assumes cross-sectional independence across all units; a second generation  CADF (cross-section augmented Dickey-Fuller) panel unit root test that assumes heterogeneous panels with cross-section dependence across all units (Barbieri 2006). The null hypothesis of a unit root is non-stationary, compared to the alternative hypothesis of a stationary series with no unit root (Barbieri 2006). The results of the IPS and Pesaran CADF unit root tests for the whole of China are listed in Table 5. The results of the unit root tests confirm that that all of the variables are non-stationary at level I(0) and stationary at their first order differences I(1). This implies that all of the variables have the same order of integration and may have a long-run cointegration relationship. Thus, a cointegration test is applied for all models in the next section.
Based on the results of the panel unit root tests shown in Table 5, it can be confirmed that all of the variables have the same order of integration and may have a long-run cointegration relationship. Thus, this study applied a Fisher-type Johansen panel cointegration test to explore the long-run equilibrium relationship among the variables of Equations (1)-(3). The Johansen cointegration test includes two unrestricted cointegration rank statistics: Trace and Maximum Eigenvalue. The results for the cointegration test in Table 6 reject the null hypothesis of no cointegration at the 5% level of significance. Thus, all variables in Equations (1)-(3) are cointegrated. 9 The environmental data are only updated to 2015 prior to the completion of this paper. (1) The symbols *** and ** indicate that the t-test is significant at the thresholds of 1% and 5%, respectively, which means that the null hypothesis is rejected.
(2) The unit root tests are estimated using constant and trend variables.
(3) All tests are undertaken using Stata software.
The cointegration test confirmed that there is a long run equilibrium relationship among these variables. Thus, the fully modified ordinary least squares (FM-OLS) method was employed on Equations (1)-(3) to explore the long-run elasticities among these variables. (Ozcan 2013;Chao et al. 2014;Farhani and Shahbaz 2014). The results obtained from the model estimated for China using the fully modified ordinary least squares (FM-OLS) method are presented in Table 7. The comparison of the expected signs of the coefficients is summarized in Tables 8 and 9, in which the actual sign is circled when it is different from the expected sign. These results suggest that GDPPC is highly significant at the 1% level in almost all of the sectors and regions, except in North East China. The elasticity of demand for water is close to one in many cases, suggesting a proportionate increase in the demand for water in response to the increase in GDPPC. The increase in population leads to an increase in water consumption for household purposes. The POP variable is significant; it has the expected positive sign in most cases of models estimated for both total and household water consumption. Exports of goods implicitly involves the export of water, particularly in the case of exports of agricultural products. The EXP variable significantly affects the demand for water, at both the aggregate and sectoral levels, and for all the regions. Similarly, the import of goods implicitly involves the import of water, particularly in the case of the import of agricultural products. The results for the IMP variable are somewhat mixed in terms of the sign and significance of its coefficient across sectors and regions. Similarly, the evidence for the effects of FDI is mixed across sectors and regions. The elasticities of water consumption with regard to FDI and IMP remain low (less than 0.5) in most cases. SO2 positively and significantly affects the demand for water. Variables Variables Variables Variables                                                                        Risk Financial Manag. 2020, 13, x FOR PEER REVIEW 20 of 27 FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively. (2) The actual sign is circled when it is different from the expected sign.  FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively. (2) The actual sign is circled when it is different from the expected sign.  FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively. (2) The actual sign is circled when it is different from the expected sign.  FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively. (2) The actual sign is circled when it is different from the expected sign.  FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively. (2) The actual sign is circled when it is different from the expected sign.  FDI +*** +*** − − + −* +* +*** Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively.

WCT
(2) The actual sign is circled when it is different from the expected sign.  Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively.
(2) The actual sign is circled when it is different from the expected sign.  Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively.
(2) The actual sign is circled when it is different from the expected sign.  Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively.
(2) The actual sign is circled when it is different from the expected sign.  Note: *, **, *** indicate that the estimated coefficients are significant at the confidence levels of 10%, 5%, 1%, respectively.

Estimates for Australia, Japan, and the UK
The panel data analysis of the demand for water in the developed economies of Australia, Japan, and the UK was carried out only at the aggregate and industrial levels because of the lack of available data for the agricultural and household sectors. The results suggest that GDPPC and EXP significantly affect the demand for water at both the aggregate and industrial levels (Tables 10 and 11). FDINI is significant, whereas the IMP and FDINO variables are insignificant in all of the equations. These results suggest that there are significant effects of GDPPC, EXP, CO2, and FDINI on the demand for water in the developed economies. Interestingly, GDPPC has a negative sign. As hypothesized earlier, the developed countries use better water saving technologies and have more efficient water management strategies. An increase in GDPPC therefore leads to a decrease in demand for water.

Conclusions
Motivated by the issue of global water scarcity, this study aimed to investigate the domestic and international drivers of the demand for water in China. China has traditionally depended on agriculture and shifted its focus to manufacturing in recent decades. China is the biggest global exporter of manufactured goods, and the largest global consumer, because of its recent economic growth and the size of its population. Analysis of the demand for water in China was carried out at the aggregate, sectoral, and provincial levels. Similar analysis of three developed countries (Australia, Japan, and the United Kingdom) provides international relevance for our study. Panel data models were estimated using annual data from 2004 to 2015 for China and from 2005 to 2013 for Australia, Japan, and the United Kingdom. The results suggest that per capita GDP is highly significant at the 1% level in most of the sectors and regions, except for North East China. The elasticity of the demand for water is close to one in many cases, suggesting a proportionate increase in the demand for water in response to an increase in per capita GDP. Variables Population (POP), total exports (EXP), and Sulfur di oxide emissions (SO2) are significant and mostly have the expected positive signs in the models estimated for both total and household water consumption. The results for the Total imports (IMP) and foreign direct investments (FDI) are mixed in terms of the coefficient signs and significance across sectors and regions. Policymakers ought to look at the specific characteristics of the region and stage of economic growth in terms of policies to appropriately manage water demand in the region. The elasticities of water consumption with regard to variables foreign direct investments (FDI) and total imports (IMP) remain low (less than 0.5) in most cases.
The results obtained for the developed economies of Australia, Japan, and the UK suggest that GDP per capita and total exports significantly affect the demand for water at both the aggregate and sectoral levels. Net inflows of foreign direct investments has significant effects on the demand for water, whereas those of total imports and net outflows of investments are insignificant. These results indicate the significant effects of per capita GDP, net exports, Sulfur dioxide emissions and net inflows of foreign direct investments on the demand for water in the developed economies. The findings of the study are useful for the formulation of relevant strategies for the efficient management of scarce water resources at both domestic and global levels. Both economic growth and globalization increase the pressure on the demand for water.
Author Contributions: Conceptualization, supervision and project administration by R.G.; Writing-original draft preparation, writing-review and editing, methodology, data curation and formal analysis by K.Y.; Review, writing and editing by T.S. Writing-original draft preparation, validation by D.M. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.