Assessing Water Resource Use Efficiency based on the Extended TwoStage Data Envelopment Analysis (Dong Nai River Basin, Vietnam)45225

Assessing Water Resource Use Efficiency based on the Extended Two-Stage Data Envelopment Analysis Dong Nai River Basin, Vietnam Nguyen Truc Le 1 , Nguyen An Thinh 1* , Nguyen Thi Vinh H

Assessing Water Resource Use Efficiency based on the Extended Two-Stage Data Envelopment Analysis (Dong Nai River Basin, Vietnam)

Nguyen Truc Le 1 , Nguyen An Thinh 1* , Nguyen Thi Vinh Ha 1 , Nguyen Dinh Tien 1 , Nguyen Duc Lam 1, Nguyen Van Hong 1 , Nguyen Tat Tuan 3 , Luc Hens 4

1 VNU-University of Economics and Business, Vietnam National University, Hanoi, Vietnam

2 VAST-Institute of Geography, Hanoi, Vietnam

3 Centre for Water Resources Warning and Forecasting (CEWAFO), Hanoi 10000, Vietnam

4 Vlaamse Instelling voor Technologisch Onderzoek (VITO), Boeretang 200, 2400 Mol, Belgium

* Correspondence: anthinhhus@gmail.com

Abstract: Population growth, economic development and changing consumption patterns in recent

years have been influenced on water resource use efficiency (WRUE) Therefore determining drivers

of WRUE requires interdisciplinary attentions including inter-sectoral and inter-regional water relationships in the economy This paper identifies the efficiency of water resource use of the Dong Nai river basin using two-stage Data Envelopment Analysis (DEA) approach that is a combination

of the regional water resource metabolic theory and the two-stage DEA model The study results show that transforming the economic development and sewage treatment both in the individual households and industry sector have affected to the WRUE in the social subsystem Sewage treatment for domestic and industrial purpose currently only reaches 55% and 66%, respectively It is necessary

to control large-scale cities in the DNRB because the large-scale cities were built during the rapid urbanization process characterized by a low level of sewage treatment Analyzing and comparing the WRUE and water resource consumption structure of representative cities shows that the shortage of water resources has negatively influenced the sustainable development of the socio-economic system

in the sub-basins of the DNRB Moreover, the socio-economic system also influences to the WRUE Therefore, the relationship between the WRUE and the socio-economic development can be analyzed using the WRUE in the social and economic subsystems This study on WRUE provides a scientific base for decision making of water resources use and management in the Dong Nai river basin

Keywords: Water Resource Use Efficiency (WRUE); Water Security; Two-Stage Data Envelopment

Analysis (DEA); Dong Nai River Basin; Vietnam

1 Introduction

Water resource use efficiency (WRUE), as a concept originates from the economy of productivity, is related to sustainable development (Guan et al., 2019) Since WRUE become

a global issue, factors affected WRUE include population growth, economic development, and changing consumption patterns (Guan et al., 2019), water pricing (Veettil et al., 2013), technological innovations, new policies, and operational improvements (Liu et al., 2016) WRUE in different regions is influenced by both environmental, management, engineering, social, and economic factors Therefore determining drivers of WRUE requires interdisciplinary considerations (Liu et al., 2016) Also inter-sectoral and inter-regional water relationships in the economy are considered Agriculture, industry, service, and other sectors are competitors for the finite water resources in integrated regions (Hsiao et al., 2007)

WRUE is measured by the volume of water consumed by a plant to produce a unit

of the output The lower the resource input requirement per unit, the higher the efficiency

(Liu et al., 2016) WUE shows economic, social, and environmental components (Karimi et al., 2012; Azad and Ancev, 2014; Liu et al., 2016) In particular WRUE in a river basin is assessed by different approaches The input-output analysis is used to establish the framework connecting water use with industrial development and quantifying WRUE (Lofting and Mcgauhey, 1968) The Multi-Sector and Multi-Factorial Logarithmic Mean Divisia Index (MLMDI) decomposition method is used to identify WUE by sector into factors order of importance (Liu et al., 2016) The comprehensive evaluation approach basing on a combination of cloud-compound fuzzy matter element-entropy models: the cloud model is used to calculate the comprehensive indicators of WRUE (Guan et al., 2019) The empirical estimation of stochastic Data Envelopment Analysis (DEA) allows analyzing the efficiency of irrigation water use in the agriculture in a basin (Veettil et al., 2012)

DEA has several advantages in analyzing WRUE: it is not necessary building a mathematical formula for the production function, which is useful in evaluating the unknown relationships of water and other input factors in the production system DEA does not rely on many data: only input and output quantities are required, while the Decision-Making Units (DMUs) are considered as “black boxes” and no information on their operation is needed for analysis DEA can be applied in the case of one or multiple inputs and one or multiple outputs It also provides a comparison opportunities for DMU efficiencies and total factor productivities However, traditional DEA only evaluates the efficiency of a system as a whole In most of researches on WRUE, the effect of water for social development is neglected In fact, water has an impact on both social and economic development aspects of an economy For example, water use does not only affect the gross domestic products (GDP) but also the growth rate of a population and its urbanization, which both reflect the dynamics of the social system This paper deals with the extended two-stage DEA approach which is a combination of the regional water resource metabolic theory and the two-stage DEA model to assess WRUE issues (Ren et al., 2016) This approach has been applied to a number of basins worldwide, such as Gansu (China) (Ren et al., 2016)

The paper is organized as follows: Section 2 presents the extended Two-stage DEA model; Section 3 introduces the water use status in the Dong Nai river basin and presents the scenario analysis and the calculation results; Conclusions and discussions are offered in Section 4

2 Methodology

2.1 Dong Nai river basin

The Vietnamese Dong Nai river basin (DNRB) has an area of over 49,600 square kilometers The basin covers the Southern Key Economic Region (SKER), 3 economic regions (Central Highland, Central Coast, and Mekong River Delta) and 9 provinces (Dong Nai, Ho Chi Minh city, Binh Duong, Binh Phuoc, Tay Ninh, Long An, Binh Thuan, Lam Dong, and Dak Nong) The Dong Nai river has an average annual flow of 1,000 m3 per second, and a total water volume of over 41 billion m3 per year The basin provides water for approximately 20 million people, over 10,000 industrial production enterprises, and nearly

70 industrial parks Because SKER acts as the most active key economic region in Vietnam, fast growing industry and urban areas in this area results in rapid increase in water

consumption demand and water pollution Also sea level rise challenges water use in the Dong Nai river basin (MONRE 2016) Figure 1 shows that 29 sub-basins of the DNRB including Ben Luc, Ben Than, Can Don, Da Nhim, Dai Ninh, Dau Tieng, Dong Nai, Dong Nai 1, Dong Nai 2, Dong Nai 3, Dong Nai 4, Dong Nai 5 , Dong Nai 6, Dong Nai 8, Dong Thap Muoi, Go Dau Ha, Ham Thuan, Nha Be, Phuoc Hoa, Saigon, Song Be, Srock Phu Mieng, Ta Pao, Tay Ninh, Vam Co Tay, Tri An, Vo Dac, Thac Mo, and Can Dang These sub-basins are defined as Decision-Making Units (DMUs), which characterize the water resource use efficiency (WRUE) of the DNRB

Figure 1 The location of 29 sub-basins in the Dong Nai river basin, Vietnam

2.2 Extended two-stage DEA model

The DEA estimates technical efficiencies of DMUs such as enterprises, banks, schools, districts, etc in using inputs to generate outputs The estimation is based on the Production Possibility Frontier (PPF) analysis PPF is used to evaluate the relative Technical Efficiencies (TE) of enterprises in an industrial sector, in which enterprises on the frontier are considered as more efficient than those under the frontier TE scores of most efficient DMUs are equal to 1 TEs of inefficient DMUs, which locate under the PPF, score below 1 The less efficient a DMU, the lower the TE score Similar to TE, the maximum values of allocative efficiency and overall efficiency are 1 DMU is most efficient when its overall efficiency is equal to 1 (Farrell 1957)

TE is measured as the output to input ratio In general, if DMUj (j = 1, 2… n) uses inputs x i (i = 1, 2,…, k) to generate outputs y r (r = 1, 2,…, m), TE j of DMUj is defined by formulas (1) to (4):

maxu,v TE j (1)

subject to:

=1 (4)

Equation (2) implies that the TE of DMUj does not exceed 1, the DMU is not outside the PPF u r and v i are the ‘price’ of output y r and input x i respectively Equation (3) shows these prices are not negative Condition (4) is applied to avoid infinite number of solutions

to input and output prices

The TE measures the efficiency at a specific time The Malmquist Total Factor Productivity (MTFP) index, which is calculated based on the DEA, is widely used to measure the change in productivity of a DMU over a period of time (Caves et al 1982; Färe

et al 1996) This index can be calculated by evaluating the distance of output vectors and input vectors during two time periods

MTFP could be decomposed into the change in technical efficiency and the technical change or change in technology

TFP change = Technical efficiency change x Technical change (5)

If the MTFP exceeds 1, there is a positive change in the total factor productivity from

period t1 to period t2 This improvement results from a positive change in technical efficiency

or in technology

The extended two-stage DEA was developed to obtain more detailed and reliable results in WRUE (Ren et al 2017) The economic development is structured in subsystems and sub-stages DEA is applied for each system or stage Two-stage DEA allows studying the process of water use, and how it runs through the different stages Operation of DMUs

is investigated in more detail instead of being considered as black boxes Based on the two-stage DEA model, the main factors impacting WRUE and economic development efficiency can be identified The relationship of water use among subsystems are also identified Therefore, this method provides policy implications for water use management and improvement of WRUE

Water is used for the whole social economic system The water resource use of the social economic system is divided into two stages (Ren et al 2017) The first stage uses water

as a direct input for the production of intermediate outputs, which are inputs to the second stage for production of the final outputs of social economic development During the first stage, the economic and the social subsystem work in parallel (Figure 2) In the social subsystem, people use water to live and for community services such as cleaning streets and public infrastructure The output of this subsystem might be measured as the population size and the percentage of urban dwellers in total In the economic subsystem, water is used

for the production of goods and services, which are measured as regional gross domestic product and the percentages of secondary and tertiary sectors

Figure 2 Water use in the socio-economic system (adapted from Ren et al., 2016)

Figure 2 shows that, x Aij (i = 1, 2, , I) and xBmj (m = 1, 2, , M) are the water inputs of

the economic and social subsystems in Stage 1A and stage 1B respectively; w kj and v nj are outputs of the two subsystems (1A and 1B) These are the inputs for the social economic

development subsystem in the Stage 2; y sj represent outputs of the social economic development subsystem in Stage 2, which are also outputs of the entire social economic system (Stages 1 + 2)

The extended two-stage DEA model is presented as follows:

- Stage 1:

Subject to α j ≤ 1 i = 1, 2,…, I (7)

λ ij , λ’ kj ≥ 0 (9)

Subject to β j ≤ 1 m = 1, 2,…, M (11)

s = (1, 2, , S)

k = (1, 2,…, K)

i = (1, 2, …, I)

DMU j

j = (1, 2, …, n)

xBmj

Stage 2

Social economic development subsystem

y sj

v nj(

w kj

Stage 1A

Economic subsystem

xAij

Stage 1B

Social subsystem

n = 1, 2,…, N (12)

mj , ’ nj ≥ 0 (13)

- Stage 2:

Subject to θ j ≤ 1 s = 1, 2,…, S (15)

µ sj ≥ 0 (17)

- Combination of two stages

max (ω1α j + ω2β j + ω3θ j) (18)

Subject to ω1 + ω2 + ω3 = 1 (19)

In which α j , β j and θ j are technical efficiencies of Stages 1A, 1B and 2 respectively; ω1,

ω2 and ω3 are weights of the three stages Restrictions α, β, θ ≤ 1 imply that all technical efficiencies of DMU j at all three stages are above 1 DMUj is most efficient if α = β = θ = 1

2.3 Data collection

29 sub-basins of the Dong Nai river basin are selected as DMUs to characterize the WRUE A series of input and output indicators are selected for both stages of the DEA Stage

1 consists of two sub-stages: a social and an economic one Most water use in the study area are input indicators of the first stage: water for domestic and public use (for the social subsystem), and water use for agriculture, industry and service sectors (economic subsystem) Output indicators of the two subsystems in the first stage include population, urban population proportion (social subsystem), Regional Gross Domestic Product (RGDP), and proportion of industry and services (economic subsystem) These indicators are selected because the social subsystem uses water mainly by residents to live As compared to the rural areas, urban areas consume more water in both quality and quantity In the economic subsystem, RGDP indicates a diversity of economic products as an output indicator The ratio between industry and service sectors has a significant influence on water use because these economic sectors use most water from the DNRB For the second stage of the two-stage DEA, the efficiency of the whole system is the focus While indicators such as population, urban population proportion, GDP, and proportion of industry and service are selected as inputs of the second stage, growth rate of population of the urban population, RGDP, and industry and service are selected as outputs Raw data of selected indicators are considered for the period of 2010-2017

Table 1 Selected inputs and outputs in a two-Stage DEA for the Dong Nai river basin, Vietnam

Stage 1A: Economic subsystem Water resource consumption (10 6

m 3 )

Population (10 3 people) Percentage of urban people (%) Stage 1B: Social subsystem Water resource consumption (10 6

m 3 )

Regional gross domestic product (RGDP) (10 6 VND) Percentage of industry and service

sectors (%) Stage 2: Social economic

development subsystem

Population (10 3 people) Percentage of urban people (%) Regional gross domestic product (RGDP) (10 6 VND) Percentage of industry and service

sectors (%)

Population growth rate (%) Urban population growth rate (%) RGDP growth rate (%) RGDP of industry and service sectors growth rate (%)

3 Results

3.1 Water resource consumption structures

Table 2 shows the water resource consumption (WRC) of the 29 sub-basins The table indicates the requirements of agriculture, livestock, aquaculture, industry, service, and others In 2017, the Dong Nai 3 and Ham Thuan sub-basins have a highest WRC (41.6 billion and 30.4 billion m3 respectively), whereas the Phuoc Hoa, Sai Gon and Can Don show the lowest values of WRC (below 5 billion m3)

Table 2 Water resource consumption (WRC) structures by sectors

of the 29 sub-basins in the period of 2010-2017 (1000 m3)

No

Sub-basins WRC

agriculture

WRC livestock

WRC aquaculture

WRC indust

ry

WRC servic

e

WR

C other

s

Total WRC

2 Ben Than 93.04 2.07 16.39 19.04 14.75 10.25 155.55

4 Da Nhim 101.31 2.26 17.85 20.73 16.07 11.16 169.37

5 Dai Ninh 66.79 1.49 11.77 13.67 10.59 7.36 111.67

6 Dau Tieng 93.86 2.05 16.18 18.79 14.56 10.11 155.55

8 Dong Nai 1 42.01 0.94 7.40 8.60 6.66 4.63 70.23

9 Dong Nai 2 112.63 2.51 19.84 23.05 17.86 12.40 188.29

10 Dong Nai 3 198.96 4.43 35.05 40.72 31.55 21.91 332.62

11 Dong Nai 4 128.34 2.86 22.61 26.26 20.35 14.13 214.56

12 Dong Nai 5 125.56 2.80 22.12 25.70 19.91 13.83 209.92

13 Dong Nai 6 112.94 2.52 19.90 23.11 17.91 12.44 188.82

14 Dong Nai 8 106.91 2.38 18.83 21.87 16.95 11.77 178.71

15 Dong Thap

Muoi

97.47 2.17 17.17 19.95 15.46 10.74 162.96

16 Go Dau Ha 37.47 0.84 6.60 7.67 5.94 4.13 62.64

17 Ham Thuan 145.64 3.25 25.66 29.81 23.10 16.04 243.49

19 Phuoc Hoa 22.39 0.41 3.21 3.72 2.88 2.00 34.61

21 Song Be 113.18 2.52 19.94 23.16 17.95 12.47 189.22

22 Srock Phu

Mieng

62.58 1.33 10.50 12.20 9.45 6.57 102.63

23 Ta Pao 114.89 2.56 20.24 23.51 18.22 12.65 192.08

24 Tay Ninh 74.15 1.65 13.06 15.17 11.76 8.17 123.96

25 Vam Co Tay 32.50 0.72 5.73 6.65 5.15 3.58 54.33

27 Vo Đac 76.79 1.72 13.53 15.72 12.18 8.55 128.49

28 Thac Mo 115.62 2.58 20.37 23.66 18.33 12.73 193.30

0

255.6

9

3891.2

0

(Source: Data from the Vietnamese provincial statistical yearbooks) During the period of 2010-2017, most water was used by agriculture (72%): planting used 60% of all water consumption Industrial production used 12% of WRC, and services 9% Only 7% of WRC was used for other applications, which include water for living and public consumption (such as schools, hospitals, transport, etc.)

3.2 WRUE in the subsystems

Table 3 presents the results of the two-stage DEA model in the Dong Nai river basin during the period 2010-2017 Values of α, β, and θ represent the relative WRUE of the social subsystem, economic subsystem, and the relative development efficiency of the socioeconomic system The WRUE of a sub-basin is the most efficient among all sub-basins when the relative WRUE equals to 1 The WRUE is increasingly inefficient compared to other sub-basins when the WRUE decreases gradually

On the WRUE in the social subsystem, Phuoc Hoa and Sai Gon scores are close to the efficiency optimum (α = 1) Water use in Can Don is technically efficient (α = 0.83), followed by Nha Be, Vam Co Tay, Can Dang, and Go Dau Ha Sub-basins of Dong Nai 1-8 show the lowest value of technical efficiency (α < 0.2) The WRUE by sub-basins for the economic subsystem is similar to the value of the social subsystem: Phuoc Hoa and Sai Gon are most efficient (β = 1) Can Don has good value of technical efficiency (β = 0.85), followed

by Nha Be, Vam Co Tay, Can Dang, and Go Dau Ha Sub-basins Dong Nai 1-8 records the lowest value of technical efficiency (β < 0.2) In the second stage, the sub-basins of Vam Co Tay, Tri An, Thac Mo and Can Dang are most technically efficient (θ = 1) Nha Be is also highly efficient with a score close to 1 (θ = 0.972) The average efficiency of the 29 sub-basins

is 0.753, indicating that the 2nd stage is generally efficient There is a distinct difference

between the social and the economic WRUE value among subsystems The value of α varies from 0.125 to 1, and the values of β vary from 0.130 to 1 The mean value of α and β of 29 sub-basins is lower than 0.5 (0.340 and 0.351 respectively), which indicates the lower efficiency in water use of the DNRB The WRUE of the economic subsystems is higher than this of the social subsystem because the mean values of α and β of 29 sub-basins is 0.351 and 0.340, respectively The difference in WRUE in the economic subsystem is more obvious than that in the social subsystem when compared to the standard deviation value of α and

β of 29 sub-basins (0.351 and 0.340 respectively) In the whole system, the sub-basins of Phuoc Hoa, Can Don and Sai Gon are most efficient (mean values of α, β and θ are 0.876, 0.805, and 0.766 respectively)

Estimating the WRUE in the DNRB, the input maximization model is selected, since water in the area is below the water demand Variable return to scale is applied to count for the differences in social economic development status of the sub-basins

Table 3 Average technical WRUE by sub-basin in the DNRB during 2010-2017

Sub-basins WRUE in social

subsystems (α)

WRUE in economic subsystems (β)

WRUE in socioeconomic subsystems (θ)

WRUE in whole socioeconomic system

Tri An 0.340 0.353 1.000 0.565

(Note: α: WRUE of social subsystem; β: WRUE economic subsystem; θ: WRUE of socioeconomic

system.)

3.2 Malmquist total factor productivity

The MTFP indices show the changes in total productivity of the 29 sub-basins Table

4 shows that the Dong Nai river basin has a positive change in total factor productivity over time All the MTFP indices of the social economic system are above 1 However, the improvement in productivity is largely due to technical changes (i.e improvement in production technology) The change mean WRUE is below 1 for all stages, which indicates that the efficiency of water use decreases gradually The technical change is most evident the economic subsystem at the Stage 1B (1.408) and lowest for the social economic subsystem

at the Stage 2 (0.995) In contrast, the efficiency change of water use is smallest at the Stage 1B and highest at the Stage 2 The rural areas in sub-basins of Dong Nai 2, 3, 4, 6 and Vam

Co Tay show most improvement in TFP while the urban areas of Ben Than, Sai Gon, Dong Nai 8, Nha Be, Phuoc Hoa have the smallest improvement in TFP This might be because the urban areas use less water than the rural areas