Decomposing the decoupling of water consumption and economic growth in Jiangxi, China

Current population growth coupled with industrial growth has caused water supply to be outstripped by human demand. Understanding water consumption (WC) decoupling patterns and the factors affecting the decoupling status are essential for balancing economic growth and WC. This study determines the decoupling relationship between WC and economic growth in Jiangxi Province, China, and the driving factors were determined by the Tapio decoupling model and the logarithmic mean Divisia index method. Results showed that changes in the industrial structure in Jiangxi Province resulted in corresponding changes in WC structure. Analysis of the decoupling relationship showed that the decoupling state between WC and economic growth for primary industry was very unstable and largely volatile from 1999 to 2015, but showed a good decoupling status for secondary and tertiary industries. The largest cumulative effects on WC were economic development and technology, which were positive and negative drivers of WC changes, contributing 1,406.14% and (cid:1) 902.96% to the total effect of WC, respectively. The ﬁ ndings can help Jiangxi government identify the key factors in ﬂ uencing the decoupling effect, and formulate effective policies to reduce WC, which will bene ﬁ t the harmonious development of economy, society and water resources in Jiangxi Province.


INTRODUCTION
Water resources are indispensable for healthy human societies, the preservation of natural environments and ecosystem services, and population and socioeconomic development. However, with rapid economic development, population growth, urbanization and climate change, problems related to water resources have posed a great threat to human security, and hindered sustainable economic development (Wu et al. ). The frequency and severity of water resource issues are both increasing, which has aroused concern among researchers worldwide (Murray et al. ; Ren et al. ). Many different methods have been proposed to understand water resources and their consumption, for example, the link between supply and demand, conflict for water resources (Böhmelt et al. ; Xiong et al. ), water footprint and water consumption in the Nile Basin (Sallam ), in Heilongjiang, China (Zhang & Yang, ), and in China (Li et al. a), water consumption and climate change (Dawadi & Ahmad ; Palazzoli et al. ), the relationship between water consumption and economic growth with structural decomposition analysis (Cazcarro et al. ), panel dataset analysis (Tir et al. ), the environmental Kuznets curve (Katz ; Shan ), and the transcendental logarithmic (translog) production model (Ngoran et al. ). These studies have shown that water consumption has a determining role in both increasing and decreasing economic growth. Although water availability is potentially able to promote economic growth, long-term economic growth can only be assured if an increase in water supply provision is identified on a sustainable basis.
Consequently, there is a need for more robust approaches to investigate water consumption and its relationship with sustainable economic growth.
Decoupling analysis is widely used and has attracted great attention in studies of economic growth in association with environmental pressures (OECD ). Decoupling refers to breaking the linkages between 'economic goods' and 'environmental bads'. Determining the decoupling relationship between the economy and the environment is a key step in achieving green economic development, which is one of the main objectives of human development as proposed by the Organisation for Economic Co-operation and Development (OECD ). Decoupling can be accomplished by compelling people to rethink the connections among resource utilization, environmental quality, and economic growth (Zhang et al. a, b). Considerable research has been conducted to establish the decoupling relationship between economic growth and water consumption. For example, Zhu et al. () studied the decoupling relationship between water use and economic development for two provinces in China (Yunnan and Guizhou), and found that the decoupling state was far from ideal. Zhang & Yang () used the water footprint method to study the decoupling relationship among water consumption, water environmental pressure, and crop production, and showed that strong decoupling occurs more often between water consumption and crop production. Gilmont () found that decoupling comprises two types in Israel: one type occurs when the economy ceases to be water self-sufficient, and the other type occurs when the economy has the capacity to remedy its over-exploitation of natural water. Jiangxi Province in China has rich water resources, but the distribution of the water resources varies largely in both time and space. Furthermore, with accelerated development of a relatively dense population and town, a relatively backward production mode, and prominent structural contradictions, as well as rapid industrial aggregation, the disparity between supply and demand of water resources in Jiangxi has become increasingly prominent, which has seriously constrained sustainable socioeconomic and regional development. On the basis of this consideration, we investigated the relationship between water consumption and economic growth in Jiangxi Province by decoupling methodology and then analyzed the driving factors of the relationship by the logarithmic mean Divisia index (LMDI), to provide a reference for the management and optimization of water resources utilization among different industries, and coordinating the relationship between economic development and water consumption. The study shows that changes in the industrial structure of Jiangxi Province resulted in corresponding changes in the water consumption structure and that economic development and technology were the main drivers of water consumption changes.

Research area
Jiangxi Province is located in southeastern China. It borders Zhejiang and Fujian in the east, Guangdong to the south, Hunan to the west, and Hubei and Anhui to the north ( Figure 1). It is a common hinterland of the Yangtze River delta, the Pearl River delta and the Hercynian economic zone. In addition, China's largest freshwater lake, Poyang Lake, is located in northern Jiangxi Province and is connected to the Yangtze River.
Jiangxi Province covers an area of 166,900 km 2 and ranges from 24 29 0 14″ to 30 04 0 41″ N and from 113 34 0 36″ to 118 28 0 58″ E. With the rapid development of the Poyang Lake ecological economic zone in recent years, Jiangxi has become the most economically active area in South China.
In particular, the Poyang Lake ecological economic zone was included in the national strategy in 2009, which has further accelerated the economic development of Jiangxi.

Tapio decoupling model
In the early 2000s, decoupling theory was introduced to social economics to study the relationships between economic growth and environmental pressures. Tapio's decoupling elasticity method (Tapio ) is used universally to describe the direction and degree of decoupling. In decoupling studies of water consumption and economic growth, the decoupling elasticity coefficient is defined as the ratio between the change rate of water consumption or environmental pressure and the change rate of economic conditions over a certain period of time. The decoupling elasticity coefficient of water consumption and economic growth is calculated by the following equation: where WC is water consumption, WC t is defined as the WC of the i industry in year t, and WC tÀ1 is defined as the WC of the i industry in year t À 1. GDP is gross domestic product, ΔWC is the change of WC, and ΔGDP is the change of GDP. ΔWC  Strong decoupling is the ideal state of decoupling, that is, a decline in water consumption with economic growth.
Strong negative decoupling is the worst situation, that is, simultaneous recession and increased water consumption.

LMDI decomposition model
The index decomposition analysis (IDA) method is widely Ang ). Thus, the LMDI model is applied in this study.
Water consumption can be calculated as: where WC i is the WC of i industry, GDP/P is the proportion of GDP to total population (P), representing the GDP per capita (g), GDP i is the GDP of i industry (in millions), year t according to the LMDI method (Ang ) can be broken down into the following formula: where ΔWC P is the contribution of total population to the annual change in WC, ΔWC g is the contribution of regional economic development to the annual change in WC, ΔWC S i is the contribution of industrial structure to the annual change in WC, and ΔWC I i is the contribution of technology to annual change in WC. The remaining terms ΔWC, ΔWC P , ΔWC g , ΔWC S i , and ΔWC I i represent the difference in WC between year t and year t À 1, for total effect, population effect, economic development effect, industrial structure effect and technology effect, respectively. The contribution of each factor based on the LMDI decomposition method (Ang ) can be expressed by the following formulas: Data source and processing   (2005)).
industry (Zhang et al. a, b), and the domestic water consumption is classified as the WC of tertiary industry

Analysis of the decoupling relationship between WC and economic growth
Decoupling analysis of total WC and total economic growth  (2001, 2002, 2003, 2006, 2008, 2010, 2012, 2014, and 2015 Decoupling analysis of WC and economic growth for primary industry  Figure 3(b)). As shown in Table 2, the GDP of primary industry continued to grow from 2000 to 2015, with a growth rate mostly around 4.5%; however, its WC was

Decoupling analysis of WC and economic growth for secondary industry
As is shown in Table 3, in terms of secondary industry, the WC growth rate was lower than its economic growth rate in all years. Expansive coupling occurred in 2000 and 2007, strong decoupling occurred in five years (2001, 2005, 2006, 2009, and 2012), and weak decoupling occurred in the remaining years (2002, 2003, 2004, 2008, 2010, 2011, 2013, 2014, and 2015).  during 2002, 2003, 2004, 2008, 2010, 2011, 2013, 2014, and 2015, there was a state of weak decoupling, with D (wc, g) values ranging from 0.036 to 0.622, which manifest as decoupling states that are generally considered satisfactory. As is shown in Figure 3(a), secondary industry has an essential role in pushing economic development, and the proportion of secondary industry increased while the proportion of the primary and tertiary industries decreased ( Figure 3(b)). This indicates that the transformation and adjustment of industrial structure has a key role in improving water-use efficiency to achieve the goal of reducing water consumption (Wu et al. ).
Decoupling analysis of WC and economic growth for tertiary industry Table 4 shows the decoupling relationship between the WC and economic growth of tertiary industry in Jiangxi   (2001, 2002, 2004, 2006, 2008, 2010, 2011, 2012, 2013, 2014, and 2015  which was the principal driver of WC increase. Water resources utilization is a basic production factor in keeping the economy running and is highly correlated with the level of economic development. The population effect played a positive role in regional WC and was a stimulative factor of WC increase, but its degree of influence was much smaller than that of the regional economic development effect.

Decomposition analysis of WC
The population effect showed a positive fluctuating trend (except for 2000) from 1999 to 2015, and its cumulative effect was 17.041 billion m 3 , accounting for 64.06% of the total effect (Table 5). Population growth is bound to be accompanied by an increase in residents' water consumption, economic scale and the water consumption, which results in an increase in water resources scarcity. During the study period, the industrial structure effect showed a downward fluctuating trend between 1999 and 2015, the cumulative effect reached À124.288 billion m 3 , and its absolute value accounted for 467.25% of the total effect (Table 5). It also was a negative driver of changes in regional WC, which indicates that industrial structure adjustment has an inhibiting effect on WC increase. As can be seen in Figure 2, the dominant industry started to shift from primary industry with high WC to second and tertiary industries with low WC, which had a positive effect on the decrease of WC.
The cumulative technology effect reached À240.187 billion m 3 , its absolute value accounted for 902.96% of the total effect, and it was a negative driver of changes in regional WC, except in 2004WC, except in , 2007WC, except in , 2009WC, except in , 2011WC, except in , and 2013 (Table 5), which indicates that technical progress has an inhibiting effect on WC increase and is the most important factor in the WC decrease ( Figure 4).
In summary, the economic development and population effects were positive drivers of WC changes, but the population effect was much smaller than the economic

CONCLUSIONS
Based on water consumption (WC) and economic growth data for the three types of industry (primary, secondary, and tertiary  industries) in Jiangxi Province from 1999 to 2015, the decoupling relationship between WC and economic growth was analyzed by the Tapio decoupling model. Furthermore, the LMDI technique was applied to identify the main driving forces affecting changes in water consumption. The main conclusions drawn from the present study are as follows.
From 1999 to 2002, the three types of industry in Jiangxi exhibited a tertiary-secondary-primary structure, and generally followed a secondary-tertiary-primary structure from 2003 to 2015. In addition, changes in the industrial structure resulted in corresponding changes in the water consumption structure.
The main factors affecting the decoupling of WC and economic growth in Jiangxi Province were the economic development level and technology level. The economic development effect was a large positive driver of WC changes and the technology effect was a large negative driver of WC changes.