留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China

Dalai MA Fengtai ZHANG Lei GAO Guangming YANG Qing YANG Youzhi AN

MA Dalai, ZHANG Fengtai, GAO Lei, YANG Guangming, YANG Qing, AN Youzhi, 2021. Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China. Chinese Geographical Science, 31(5): 795−814 doi:  10.1007/s11769-021-1227-3
Citation: MA Dalai, ZHANG Fengtai, GAO Lei, YANG Guangming, YANG Qing, AN Youzhi, 2021. Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China. Chinese Geographical Science, 31(5): 795−814 doi:  10.1007/s11769-021-1227-3

doi: 10.1007/s11769-021-1227-3

Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China

Funds: Under the auspices of Chinese Ministry of Education Humanities and Social Sciences Project (No. 19YJCZH241), Project of Chongqing Social Science Planning Project of China (No. 2020QNGL38), Science and Technology Research Program of Chongqing Education Commission of China (No. KJQN201901143), Humanities and Social Sciences Research Program of Chongqing Education Commission of China (No. 20SKGH169)
More Information
    • 关键词:
    •  / 
    •  / 
    •  / 
    •  / 
    •  
  • Figure  1.  Location and geographical divisions of China. Tibet, Hong Kong, Macao and Taiwan of China are not included as no data

    Figure  2.  Green total-factor water-use efficiencies of different regions of China from 2000 to 2018

    Figure  3.  Local indication of spatial association maps of provincial green total-factor water-use efficiencies in China in 2000 and 2018

    Table  1.   Statistical description of the input and output variables for measuring green total-factor water-use efficiency (GTFWUE) in China

    VariableUnitObs.MinMaxMeanSD
    Labor force 104 person 570 275.50 6766.00 2530.04 1680.30
    Capital stock 108 yuan (RMB) 570 211.80 192708.40 30437.67 36245.90
    Water consumption 108 m3 570 19.96 592.00 194.04 136.59
    Economic value 108 yuan (RMB) 570 263.59 58545.69 10201.68 10373.25
    Wastewater discharge 108 t 570 1.10 93.83 19.68 15.98
    Notes: Obs. is the number of observation samples. SD is the standard deviation
    下载: 导出CSV

    Table  2.   Statistical description of factors influencing on green total-factor water-use efficiency (GTFWUE) in China

    VariablesUnitObs.MinMaxMeanSD
    Economic Growth (EG) yuan (RMB) 570 7.92 11.85 10.08 0.85
    Population Size (PS) 104 person 570 6.25 9.34 8.16 0.76
    Water Endowment (WE) m3/person 570 5.08 7.89 6.04 0.57
    Water-use Structure (WS) % 570 10.69 95.19 61.51 17.39
    Technological Progress (TP) 108 yuan (RMB) 570 1.95 13.08 8.72 1.88
    Opening-up Level (OL) % 570 0.04 15.36 2.55 2.23
    Government Influence (GI) % 570 0.42 18.97 7.77 4.94
    Urbanization Level (UL) % 570 23.20 89.60 49.96 15.13
    下载: 导出CSV

    Table  3.   Green total-factor water-use efficiencies (GTFWUEs) of Chinese provinces during 2000–2018

    Province2000200220042006200820102012201420162018Average
    Tianjin 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
    Liaoning 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
    Shanghai 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
    Yunnan 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
    Fujian 1.0000 1.0000 0.7680 0.8872 0.8589 0.8646 0.7950 0.8108 0.8230 0.8582 0.8605
    Shandong 1.0000 1.0000 0.7779 0.7328 0.7848 0.7824 0.7406 0.7777 0.7924 0.8284 0.8263
    Zhejiang 0.9795 0.8774 0.7741 0.8357 0.8162 0.7860 0.7635 0.7942 0.8090 0.7706 0.8150
    Heilongjiang 0.8095 0.8120 0.7770 1.0000 1.0000 1.0000 0.6315 0.6186 0.6188 0.6232 0.8064
    Anhui 0.8033 0.8474 0.8867 0.8794 0.8658 0.7624 0.7803 0.8085 0.7096 0.6959 0.7996
    Sichuan 0.7429 0.7570 0.7647 0.7932 0.6878 0.7393 0.7869 0.8151 0.8262 0.8419 0.7745
    Hebei 0.9061 0.8659 0.8459 0.8609 0.7609 0.7436 0.6662 0.6861 0.6709 0.6367 0.7647
    Hubei 0.8162 0.7975 0.7712 0.8339 0.7578 0.7742 0.6883 0.7025 0.6880 0.6993 0.7575
    Guangdong 0.8562 0.7259 0.6689 0.7927 0.7767 0.7657 0.6513 0.6556 0.6580 0.6678 0.7179
    Chongqing 0.6436 0.6871 0.7056 0.7237 0.6530 0.7135 0.7356 0.7773 0.7249 0.7099 0.7119
    Jilin 0.8323 0.7956 0.7388 0.8088 0.7589 0.7568 0.5859 0.5979 0.5568 0.6219 0.7097
    Beijing 1.0000 0.7479 0.6607 0.6638 0.6740 0.6608 0.6769 0.6811 0.6376 0.7014 0.7089
    Hainan 0.7030 0.7572 0.6536 0.7906 0.7730 0.7239 0.6432 0.6507 0.6479 0.6663 0.7062
    Hunan 0.7553 0.7502 0.6926 0.7860 0.7414 0.6948 0.6386 0.6473 0.6678 0.6507 0.7055
    Jiangsu 0.9013 0.6978 0.6156 0.7424 0.7422 0.7538 0.6210 0.6347 0.6352 0.6489 0.6952
    Shanxi 0.7561 0.7550 0.7187 0.7209 0.6628 0.6283 0.5958 0.5985 0.5762 0.5969 0.6601
    Henan 0.8133 0.7603 0.7206 0.7497 0.6567 0.6240 0.5666 0.5809 0.5633 0.5803 0.6563
    Qinghai 0.6017 0.7136 0.6638 0.7222 0.6798 0.6494 0.5837 0.5830 0.5622 0.5722 0.6360
    Shaanxi 0.7970 0.7434 0.6321 0.6366 0.5642 0.5740 0.6000 0.6113 0.5853 0.5751 0.6286
    Guizhou 0.6461 0.6690 0.5771 0.6078 0.5856 0.6300 0.6398 0.6500 0.6010 0.6808 0.6260
    Inner Mongolia 0.7380 0.6567 0.5759 0.4898 0.6711 0.6936 0.4558 0.4553 0.4320 0.4870 0.5706
    Guangxi 0.6029 0.5889 0.5714 0.6252 0.6136 0.5711 0.4787 0.4989 0.5606 0.5492 0.5690
    Jiangxi 0.7373 0.5737 0.5182 0.5975 0.5913 0.5496 0.4954 0.4861 0.5083 0.5504 0.5560
    Xinjiang 0.6143 0.5940 0.5326 0.5434 0.5266 0.5228 0.4672 0.4577 0.4306 0.4299 0.5149
    Gansu 0.6125 0.5105 0.5358 0.5721 0.5770 0.5599 0.4263 0.4516 0.4323 0.4291 0.5117
    Ningxia 0.5170 0.4951 0.4647 0.4851 0.4684 0.4766 0.3758 0.3882 0.4045 0.4013 0.4524
    Note: Due to space limitations, only the green total-factor water-use efficiency values for the even-numbered years are listed. Not including Tibet, Hong Kong, Macao and Taiwan of China
    下载: 导出CSV

    Table  4.   Global Moran’s I of green total-factor water-use efficiencies (GTFWUEs) in China

    YearMoran’s IE(I)MeanSD(I)Z(I)
    20000.4002–0.0345–0.04010.11913.6499***
    20010.2048–0.0345–0.03680.12321.9424***
    20020.1306–0.0345–0.03650.12451.3261*
    20030.1298–0.0345–0.03800.12621.3019*
    20040.0624–0.0345–0.04010.11770.8233
    20050.1043–0.0345–0.04270.12151.1424*
    20060.1538–0.0345–0.03790.11981.5718*
    20070.1876–0.0345–0.04020.12021.8478**
    20080.1176–0.0345–0.03830.12521.2149*
    20090.1899–0.0345–0.03860.11881.8889**
    20100.1021–0.0345–0.04650.12511.0919*
    20110.0780–0.0345–0.03720.12170.9244
    20120.1066–0.0345–0.03950.11931.1827*
    20130.0873–0.0345–0.04470.12181.0000*
    20140.1259–0.0345–0.03660.12061.3300*
    20150.1568–0.0345–0.03660.12111.5797*
    20160.1318–0.0345–0.03900.11971.3893*
    20170.1228–0.0345–0.04390.11921.3196*
    20180.1632–0.0345–0.04130.11561.7102**
    Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. The explanation of E(I), SD(I) and Z(I) can be found in Equations (6)–(8)
    下载: 导出CSV

    Table  5.   Estimation results of common panel data models

    VariablesNon fixed effectSpace fixed effectTime fixed effectTwo-way fixed effect
    Economic Growth (EG) 0.0205
    (1.1245)
    −0.0570***
    (−3.8886)
    0.0643**
    (2.3651)
    0.0246
    (0.9509)
    Population Size (PS) 0.1281***
    (11.3144)
    0.0240
    (0.3578)
    0.1313***
    (12.1319)
    0.0490
    (0.6928)
    Water Endowment (WE) −0.0536***
    (−5.3478)
    0.0328
    (0.8948)
    −0.0484***
    (−5.0122)
    0.0076
    (0.2285)
    Water-use Structure(WS) −0.0198
    (−0.4411)
    0.0170
    (0.2382)
    −0.0071
    (−0.1581)
    0.0539
    (0.0539)
    Technological Progress (TP) −0.0632***
    (−9.9968)
    −0.0020
    (−0.3104)
    −0.0708***
    (−11.4664)
    −0.0055
    (−0.7974)
    Opening-up Level (OL) 2.0881***
    (7.7781)
    0.2567
    (1.1345)
    1.8937***
    (6.7982)
    0.2301
    (1.1567)
    Government Influence (GI) −0.5671***
    (−3.5262)
    −0.2297**
    (−2.2727)
    −1.3898***
    (−4.9916)
    −0.7290***
    (−4.3731)
    Urbanization Level (UL) 0.6466***
    (6.9222)
    0.2701***
    (2.6005)
    0.5009***
    (5.0394)
    0.2384***
    (2.5729)
    R2 0.4878 0.2360 0.5096 0.0513
    Log-L 421.7723 787.1394 455.7593 867.5175
    DW 2.0404 1.3724 2.1387 1.8060
    LM-lag 0.0003 139.9561*** 5.6365** 27.2893***
    Robust LM-lag 3.1278* 2.4024* 0.9824 1.0044
    LM-err 0.9373 145.3999*** 4.7146** 29.2830***
    Robust LM-err 4.0648** 7.8461*** 0.0605 2.9981*
    Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. The Robust LM-lag is the robustness test on Lagrange multiplier (LM) of spatial lag, while Robust LM-err is the robustness test on LM of spatial err. If the statistic value of the Robust LM-lag is greater than that of Robust LM-err, the spatial autoregressive model has better explanatory power. On the contrary, spatial error model has more advantages in variable interpretation
    下载: 导出CSV

    Table  6.   Estimation results of time fixed effect spatial econometric models

    VariablesSpatial autoregressive modelSpatial error model
    Economic Growth (EG) 0.0749***
    (2.7931)
    0.0751***
    (2.8254)
    Population Size (PS) 0.1343***
    (12.6068)
    0.1358***
    (12.8389)
    Water Endowment (WE) −0.0520***
    (−5.3816)
    −0.0430***
    (−4.7091)
    Water-use Structure (WS) −0.0271
    (−0.6165)
    −0.0460
    (−1.0886)
    Technological Progress (TP) −0.0708***
    (−11.6075)
    −0.0714***
    (−11.6439)
    Opening-up Level (OL) 1.9371***
    (7.0073)
    1.7588***
    (6.4369)
    Government Influence (GI) −1.3302***
    (−4.8535)
    −1.2172***
    (−4.5878)
    Urbanization Level (UL) 0.4904***
    (5.0112)
    0.4865***
    (5.0866)
    W× dep. var. −0.1499***
    (−3.0767)
    spat. aut. −0.1640***
    (−2.7519)
    R2 0.5533 0.5439
    Log-L 459.0934 458.8562
    Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. W × dep. var.is the spatial lag term of the spatial autoregressive model, while spat. aut.is the spatial error term of the spatial error mode. Log-L is the log likelihood function
    下载: 导出CSV
  • [1] Acemoglu D, Aghion P, Bursztyn L et al., 2012. The environment and directed technical change. American Economic Review, 102(1): 131–166. doi:  10.1093/oxrep/gru031
    [2] Anselin L, 1988. Spatial Econometrics: Methods and Models. Dordrecht: Springer Science.
    [3] Anselin L, Getis A, 1992. Spatial statistical analysis and geographic information systems. The Annals of Regional Science, 26(1): 19–33. doi:  10.1007/BF01581478
    [4] Anselin L, Syabri I, Kho Y, 2010. GeoDa: An Introduction to Spatial Data Analysis: Handbook of Applied Spatial Analysis. Berlin: Springer, 73–89.
    [5] Azad A S, Ancev T, Hernández-Sancho F, 2015. Efficient water use for sustainable irrigation industry. Water Resources Management, 29(5): 1683–1696. doi:  10.1007/s11269-014-0904-8
    [6] Bryan B A, Gao L, Ye Y et al., 2018. China’s response to a national land-system sustainability emergency. Nature, 559(7713): 193–204. doi:  10.1038/s41586-018-0280-2
    [7] Cao X C, Ren J, Wu M Y et al., 2018. Effective use rate of generalized water resources assessment and to improve agricultural water use efficiency evaluation index system. Ecological Indicators, 86: 58–66. doi:  10.1016/j.ecolind.2017.12.016
    [8] Chen Q, Ai H S, Zhang Y B et al., 2019. Marketization and water resource utilization efficiency in China. Sustainable Computing: Informatics and Systems, 22: 32–43. doi:  10.1016/j.suscom.2019.01.018
    [9] Chen S J, Cao Y Y, Li J, 2021. The effect of water rights trading policy on water resource utilization efficiency: evidence from a quasi-natural experiment in China. Sustainability, 13(9): 5281. doi:  10.3390/su13095281
    [10] Chemak F, Boussemart J P, Jacquet F, 2010. Farming system performance and water use efficiency in the Tunisian semi‐arid region: data envelopment analysis approach. International Transactions in Operational Research, 17(3): 381–396. doi:  10.1111/j.1475-3995.2009.00736.x
    [11] Deason J P, Schad T M, Sherk G W, 2001. Water policy in the United States: a perspective. Water Policy, 3(3): 175–192. doi:  10.1016/S1366-7017(01)00011-3
    [12] Deng G Y, Li L, Song Y N, 2016. Provincial water use efficiency measurement and factor analysis in China: based on SBM-DEA model. Ecological Indicators, 69: 12–18. doi:  10.1016/j.ecolind.2016.03.052
    [13] Ding X L, Fu Z, Jia H W, 2019. Study on urbanization level, urban primacy and industrial water utilization efficiency in the Yangtze River Economic Belt. Sustainability, 11(23): 6571. doi:  10.3390/su11236571
    [14] Ding Xuhui, He Juhua, Wang Liuyuan, 2018. Inter-provincial water resources utilization efficiency and its driving factors considering undesirable outputs: based on SE-SBM and Tobit model. China Population, Resources and Environment, 28(1): 157–164. (in Chinese)
    [15] Dong H J, GengY, Fujita T et al., 2014. Uncovering regional disparity of China’s water footprint and inter-provincial virtual water flows. Science of the total environment, 500–501: 120–130. doi:  10.1016/j.scitotenv.2014.08.094
    [16] Elhorst J P, 2010. Spatial panel data models. In: Fischer M M and Getis A (eds.). Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications. Berlin: Springer Berlin Heidelberg, 377–407.
    [17] Fischer M M, Getis A, 2009. Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications. Berlin: Springer.
    [18] Fujii H, Managi S, Kaneko S, 2013. Decomposition analysis of air pollution abatement in China: empirical study for ten industrial sectors from 1998 to 2009. Journal of Cleaner Production, 59: 22–31. doi:  10.1016/j.jclepro.2013.06.059
    [19] Gao L, Bryan B A, 2017. Finding pathways to national-scale land-sector sustainability. Nature, 544(7649): 217–222. doi:  10.1038/nature21694
    [20] Geng Q L, Ren Q, Nolan R H et al., 2019. China’s agricultural water use efficiency in a green-blue water perspective: a study based on data envelopment analysis. Ecological Indicators, 86: 58–66. doi:  10.1016/j.ecolind.2018.09.011
    [21] Guo Zheng, Sophia Shuang Chen, Yao Shimou et al., 2021. Does foreign direct investment affect SO2 emissions in the Yangtze River Delta? A spatial econometric analysis. Chinese Geographical Science, 31(3): 400–412. doi:  10.1007/s11769-021-1197-5
    [22] Haining R, 1993. Spatial Data Analysis in the Social and Environmental Sciences. Cambridge: Cambridge University Press.
    [23] Hu J L, Wang S C, Yeh F Y, 2006. Total-factor water efficiency of regions in China. Resources Policy, 31(4): 217–230. doi:  10.1016/j.resourpol.2007.02.001
    [24] Jahanshahloo G R, Vakili J, Zarepisheh M, 2012. A linear bilevel programming problem for obtaining the closest targets and minimum distance of a unit from the strong efficient frontier. Asia-Pacific Journal of Operational Research, 29(2): 1250011. doi:  10.1142/S021759591250011X
    [25] Jefferson G H, Bai H M, Guan X J, 2006. R&D performance in Chinese industry. Economics of Innovation New Technology, 15(4–5): 345–366. doi:  10.1080/10438590500512851
    [26] Kaneko S, Tanaka K, Toyota T et al., 2004. Water efficiency of agricultural production in China: regional comparison from 1999 to 2002. International Journal of Agricultural Resources, Governance and Ecology (IJARGE), 3(3–4): 231–251. doi:  10.1504/IJARGE.2004.006038
    [27] Kurle D, Thiede S, Herrmann C, 2015. A tool-supported approach towards water efficiency in manufacturing. Procedia Cirp, 28: 34–39. doi:  10.1016/j.procir.2015.04.007
    [28] Li K, Lin B Q, 2015. Measuring green productivity growth of Chinese industrial sectors during 1998–2011. China Economic Review, 36: 279–295. doi:  10.1016/j.chieco.2015.09.008
    [29] Liang Jingxi, Zhang Ankang, Li Caifeng, 2018. An empirical study on agriculture irrigation water use efficiency based on weight constraints DEA and Tobit model—a case study of Heilongjiang Province. Water Saving Irrigation, (4): 62–68. (in Chinese)
    [30] Liu K D, Yang, G L, Yang D G, 2020. Industrial water-use efficiency in China: regional heterogeneity and incentives identification. Journal of Cleaner Production, 258: 120828. doi:  10.1016/j.jclepro.2020.120828
    [31] Ma H L, Shi C L, Chou N T, 2016. China’s water utilization efficiency: an analysis with environmental considerations. Sustainability, 8(6): 516. doi:  10.3390/su8060516
    [32] Mei K, Liao L L, Zhu Y L et al., 2014. Evaluation of spatial-temporal variations and trends in surface water quality across a rural-suburban-urban interface. Environmental Science and Pollution Research, 21(13): 8036–8051. doi:  10.1007/s11356-014-2716-z
    [33] Mielnik O, Goldemberg J, 2002. Foreign direct investment and decoupling between energy and gross domestic product in developing countries. Energy Policy, 30(2): 87–89. doi:  10.1016/S0301-4215(01)00080-5
    [34] Molinos-Senante M, Maziotis A, Sala-Garrido R, 2014. The Luenberger productivity indicator in the water industry: an empirical analysis for England and Wales. Utilities Policy, 30: 18–28. doi:  10.1016/j.jup.2014.07.001
    [35] Moran P A P, 1948. The interpretation of statistical maps. Journal of the Royal Statistical Society: Series B (Methodological), 10(2): 243–251. doi:  10.1111/j.2517-6161.1948.tb00012.x
    [36] Moran P A P, 1950. Notes on continuous stochastic phenomena. Biometrika, 37(1–2): 17–23. doi:  10.2307/2332142
    [37] National Bureau of Statistics of China (NBSC), 2019. China Statistical Yearbook. Beijing: China Statistics Press. (in Chinese)
    [38] National Bureau of Statistics of China (NBSC), 2001a−2019a. China Statistical Yearbook. Beijing: China Statistics Press. (in Chinese)
    [39] National Bureau of Statistics of China (NBSC), 2001b−2019b. China Statistical Yearbook on Science and Technology. Beijing: China Statistics Press. (in Chinese)
    [40] National Bureau of Statistics of China (NBSC), 2001c−2019c. China Population and Employment Statistics Yearbook. Beijing: China Statistics Press. (in Chinese)
    [41] National Bureau of Statistics of China (NBSC), Ministry of Ecology and Environment (MEE), 2001−2019. China Statistics Yearbook on Environment. Beijing: China Statistics Press. (in Chinese)
    [42] Njiraini G W, Guthiga P M, 2013. Are small-scale irrigators water use efficient? Evidence from Lake Naivasha Basin, Kenya. Environmental Management, 52(5): 1192–1201. doi:  10.1007/s00267-013-0146-1
    [43] Qi Q, Song S B, 2020. Measurement and influencing factors of industrial water resource utilization efficiency in Yangtze River Economic Belt. International Journal of Design & Nature and Ecodynamics, 15(5): 653–658. doi:  10.18280/ijdne.150506
    [44] Qian Wenjing, He Canfei, 2011. China’s regional difference of water resource use efficiency and influencing factors. China Population, Resources and Environment, 21(2): 54–60. (in Chinese)
    [45] Sahin O, Stewart R A, Helfer F, 2015. Bridging the water supply-demand gap in Australia: coupling water demand efficiency with rain-independent desalination supply. Water Resources Management, 29(2): 253–272. doi:  10.1007/s11269-014-0794-9
    [46] Shan Haojie, 2008. Re-estimating the capital stock of China: 1952−2006. The Journal of Quantitative & Technical Economics, 25(10): 17–31. (in Chinese)
    [47] Shi Tiange, Zhang Xiaole, Du Hongru et al., 2015. Urban water resource utilization efficiency in China. Chinese Geographical Science, 25(6): 684–697. doi:  10.1007/s11769-015-0773-y
    [48] Song M L, Wang R, Zeng X Q, 2018. Water resources utilization efficiency and influence factors under environmental restrictions. Journal of Cleaner Production, 184: 611–621. doi:  10.1016/j.jclepro.2018.02.259
    [49] Sun Caizhi, Ma Qifei, Zhao Liangshi, 2020. Analysis of driving mechanism based on a GWR model of green efficiency of water resources in China. Acta Geographica Sinica, 75(5): 1022–1035. (in Chinese)
    [50] Sun C Z, Zhao L S, Zou W et al., 2014a. Water resource utilization efficiency and spatial spillover effects in China. Journal of Geographical Sciences, 24(5): 771–788. doi:  10.1007/s11442-014-1119-x
    [51] Sun Caizhi, Zhao Liangshi, Zou Wei, 2014b. The interprovincial water resources global environmental technology efficiency measurement in China and its spatial effect. Journal of Natural Resources, 29(4): 553–563. (in Chinese)
    [52] Tone K, 2001. A slacks-based measure of efficiency in data envelopment analysis. European journal of operational research, 130(3): 498–509. doi:  10.1016/S0377-2217(99)00407-5
    [53] Wang F T, Yu C, Xiong L C et al., 2019. How can agricultural water use efficiency be promoted in China? A spatial-temporal analysis. Resources, Conservation and Recycling, 145: 411–418. doi:  10.1016/j.resconrec.2019.03.017
    [54] Wang G F, Chen J C, Wu F et al., 2015. An integrated analysis of agricultural water-use efficiency: a case study in the Heihe River Basin in Northwest China. Physics and Chemistry of the Earth, Parts A/B/C, 89–90: 3–9. doi:  10.1016/j.pce.2015.10.009
    [55] Wang L, Chen Z C, Ma D L et al., 2013. Measuring carbon emissions performance in 123 Countries: application of minimum distance to the strong efficiency frontier analysis. Sustainability, 5(12): 5319–5332. doi:  10.3390/su5125319
    [56] Wang M Q, Huang Y, Li D, 2021. Assessing the performance of industrial water resource utilization systems in China based on a two-stage DEA approach with game cross efficiency. Journal of Cleaner Production, 312: 127722. doi:  10.1016/j.jclepro.2021.127722
    [57] Wang W P, Gao L, Liu P et al., 2014. Relationships between regional economic sectors and water use in a water-scarce area in China: a quantitative analysis. Journal of Hydrology, 515: 180–190. doi:  10.1016/j.jhydrol.2014.04.057
    [58] Wang Y S, Bian Y W, Xu H, 2015. Water use efficiency and related pollutants’ abatement costs of regional industrial systems in China: a slacks-based measure approach. Journal of Cleaner Production, 101: 301–310. doi:  10.1016/j.jclepro.2015.03.092
    [59] Wei J, Zhang H Q, Liu S S et al., 2019. Technological innovation, environmental regulation, and green total factor efficiency of industrial water resources. Journal of Cleaner Production, 211: 61–69. doi:  10.1016/j.jclepro.2018.11.172
    [60] Wibowo A, Alfen H W, 2015. Benchmarking the efficiencies of Indonesia’s municipal water utilities using Stackelberg data envelopment analysis. Benchmarking: An International Journal, 22(4): 588–609. doi:  10.1108/BIJ-01-2014-0009
    [61] Wolfe J R, Goldstein R A, Maulbetsch J S et al., 2009. An electric power industry perspective on water use efficiency. Journal of Contemporary Water Research & Education, 143(1): 30–34. doi:  10.1111/j.1936-704X.2009.00062.x
    [62] Xie H L, Chen Q R, Lu F C et al., 2019. Spatial-temporal disparities and influencing factors of total-factor green use efficiency of industrial land in China. Journal of Cleaner Production, 207: 1047–1058. doi:  10.1016/j.jclepro.2018.10.087
    [63] Xu S C, Liang H, 2020. Study of the utilization efficiency and its influencing factors of water resources in Yangtze River Economic Belt. Journal of Physics: Conference Series, 1549: 022017. doi:  10.1088/1742-6596/1549/2/022017
    [64] Yao X L, Feng W, Zhang X L et al., 2018. Measurement and decomposition of industrial green total factor water efficiency in China. Journal of Cleaner Production, 198: 1144–1156. doi:  10.1016/j.jclepro.2018.07.138
    [65] Yang F, Wang D W, Zhao L L et al., 2021. Efficiency evaluation for regional industrial water use and wastewater treatment systems in China: a dynamic interactive network slacks-based measure model. Journal of Environmental Management, 279: 111721. doi:  10.1016/j.jenvman.2020.111721
    [66] Yang G M, Zhang F T, Zhang F et al., 2021. Spatiotemporal changes in efficiency and influencing factors of China’s industrial carbon emissions. Environmental Science and Pollution Research, 28: 36288–36302. doi:  10.1007/s11356-021-13003-8
    [67] Yang J, Liu X J, Ying L M et al., 2019. Correlation analysis of environmental treatment, sewage treatment and water supply efficiency in China. Science of The Total Environment, 708(1): 135128. doi:  10.1016/j.scitotenv.2019.135128
    [68] Yang Jiangmin, Tan Yiming, Xue Desheng et al., 2021. The environmental impacts of informal economies in China: inverted U-shaped relationship and regional variances. Chinese Geographical Science, 31(4): 585–599. doi:  10.1007/s11769-021-1210-z
    [69] Yang Yang, Jiang Shubin, 2016. Irrigation water efficiency evaluation of agricultural irrigation water efficiency based on DEA and Malmquist index. Ecological Economy, 32(5): 147–151. (in Chinese)
    [70] Zeng P H, Wei X C, 2021. Measurement and convergence of transportation industry total factor energy efficiency in China. Alexandria Engineering Journal, 60(5): 4267–4274. doi:  10.1016/j.aej.2021.03.032
    [71] Zhang Lingling, Ding Xueli, Shen Ying et al., 2019. Spatial heterogeneity and influencing factors of agricultural water use efficiency in China. Resources and Environment in the Yangtze Basin, 28(4): 817–827. (in Chinese)
    [72] Zhang X L, Kong Y S, Ding X H, 2020. How high-quality urbanization affects utilization efficiency of agricultural water resources in the Yellow River Basin under double control action? Sustainability, 12(7): 2869. doi:  10.3390/su12072869
    [73] Zhang Xiyue, Sun Fangcheng, Wang Huaizu, 2020. A study of the green efficiency of industrial water resource in Yangtze River Economic Zone based on the Two-stage evaluation of production and treatment. Journal of Jiangxi University of Finance and Economics, (2): 26–36. (in Chinese)
    [74] Zhao H P, Wang Z F, Chan T C et al., 2016. Design of regeneration recycling water networks by means of concentration potentials and a linear programming method. Journal of Cleaner Production, 112: 4667–4673. doi:  10.1016/j.jclepro.2015.07.130
    [75] Zhao Jing, Ni Hongzhen, Chen Genfa, 2015. Evaluation on water use efficiency of water-intensive industries in China. Water Resources and Hydropower Engineering, 46(4): 11–15. (in Chinese)
    [76] Zhao L S, Sun C Z, Liu F C, 2017a. Interprovincial two-stage water resource utilization efficiency under environmental constraint and spatial spillover effects in China. Journal of Cleaner Production, 164: 715–725. doi:  10.1016/j.jclepro.2017.06.252
    [77] Zhao Liangshi, Sun Caizhi, Liu Fengchao, 2017b. Two-tage utilization efficiency of the interprovincial water resources under environmental constraint and its influence factors in China. China Population, Resource and Environment, 27(5): 27–36. (in Chinese)
    [78] Zhao Xueyan, Wang Rong, Wang Xiaoqi et al., 2019. Spatio-temporal distribution and influencing factors of environmental pollution incidents based on multi-scales in China. Scientia Geographica Sinica, 39(9): 1361–1370. (in Chinese)
    [79] Zhou Liang, Zhou Chenhu, Yang Fan et al., 2017. Spatio-temporal evolution and the influencing factors of PM2.5 in China between 2000 and 2011. Acta Geographica Sinica, 72(11): 2079–2092. (in Chinese)
    [80] Zhou Z X, Wu H Q, Song P F, 2019. Measuring the resource and environmental effciency of industrial water consumption in China: a non-radial directional distance function. Journal of Cleaner Production, 240: 118169. doi:  10.1016/j.jclepro.2019.118169
    [81] Zoebl D, 2006. Is water productivity a useful concept in agricultural water management? Agricultural Water Management, 84(3): 265–273. doi:  10.1016/j.agwat.2006.03.002
    [82] Zou D L, Cong H B, 2020. Evaluation and influencing factors of China’s industrial water resource utilization efficiency from the perspective of spatial effect. Alexandria Engineering Journal, 60(1): 173–182. doi:  10.1016/j.aej.2020.06.053
  • [1] Yue WANG, Yuting JIANG, Guoxu ZHU.  Spatio-temporal Evaluation of Multi-scale Cultivated Land System Resilience in Black Soil Region from 2000 to 2019: A Case Study of Liaoning Province, Northeast China . Chinese Geographical Science, 2024, 34(1): 168-180. doi: 10.1007/s11769-023-1405-6
    [2] Yongwang CAO, Rongrong ZHANG, Dahao ZHANG, Chunshan ZHOU.  Urban Agglomerations in China: Characteristics and Influencing Factors of Population Agglomeration . Chinese Geographical Science, 2023, 33(4): 719-735. doi: 10.1007/s11769-023-1368-7
    [3] Jie XIAO, Jiajun QIAO, Dong HAN, Yang LIU, Tao PAN.  Spatial Coupling Relationship and Driving Mechanism of Population and Economy in Rural Areas in Qinling-Daba Mountains, China . Chinese Geographical Science, 2023, 33(5): 779-795. doi: 10.1007/s11769-023-1379-4
    [4] Yanwen SHI, Xiaojian LI, Xueyao HU, Zeyi LI.  Spatiotemporal Evolution and Influencing Factors of Urban-rural Construction Land in Rural Industrialized Areas in China: Case Studies in Changyuan City and Xinxiang County of Henan Province . Chinese Geographical Science, 2023, 33(5): 850-864. doi: 10.1007/s11769-023-1383-8
    [5] Guojun ZHANG, Yu CHEN, Gengnan WANG, Chunshan ZHOU.  Spatial-temporal Evolution and Influencing Factors of Digital Financial Inclusion: County-level Evidence from China . Chinese Geographical Science, 2023, 33(2): 221-232. doi: 10.1007/s11769-023-1333-5
    [6] Erling LI, Mengzhen ZHANG, Ruolan LI, Qingqing DENG.  Influencing Factors and Improvement Suggestions for Agricultural Green Development Performance: Empirical Insights from China . Chinese Geographical Science, 2023, 33(5): 917-933. doi: 10.1007/s11769-023-1385-6
    [7] Yi LI, Bin FANG, Yurui LI, Weilun FENG, Xu YIN.  Spatiotemporal Pattern of Cultivated Land Pressure and Its Influencing Factors in the Huaihai Economic Zone, China . Chinese Geographical Science, 2023, 33(2): 287-303. doi: 10.1007/s11769-023-1334-4
    [8] Haipeng ZHANG, Hanchu LIU, Yong SUN, Renwei HE.  Spatial Differentiation Characteristics of Human Settlements and Their Responses to Natural and Socioeconomic Conditions in the Marginal Zone of an Uninhabited Area, Changtang Plateau, China . Chinese Geographical Science, 2022, 32(3): 506-520. doi: 10.1007/s11769-022-1280-6
    [9] Ruiling HAN, Lingling LI, Xiaoyan ZHANG, Zi LU, Shaohua ZHU.  Spatial-temporal Evolution Characteristics and Decoupling Analysis of Influencing Factors of China’s Aviation Carbon Emissions . Chinese Geographical Science, 2022, 32(2): 218-236. doi: 10.1007/s11769-021-1247-z
    [10] Hu YU, Xiaoyao ZHANG, Yu DENG.  Spatiotemporal Evolution and Influencing Factors of Landscape Ecological Vulnerability in the Three-River-Source National Park Region . Chinese Geographical Science, 2022, 32(5): 852-866. doi: 10.1007/s11769-022-1297-x
    [11] Xingchuan GAO, Tao LI, Dongqi SUN.  Regional Differentiation Regularity and Influencing Factors of Population Change in the Qinghai-Tibet Plateau, China . Chinese Geographical Science, 2021, 31(5): 888-899. doi: 10.1007/s11769-021-1223-7
    [12] Zheng CAO, Ya WEN, Song SONG, Chak Ho HUNG, Hui SUN.  Spatiotemporal Variations and Controls on Anthropogenic Heat Fluxes in 12 Selected Cities in the Eastern China . Chinese Geographical Science, 2021, 31(3): 444-458. doi: 10.1007/s11769-021-1203-y
    [13] Wenwen QIU, Zhangbao ZHONG, Zhaoliang LI.  Agricultural Non-point Source Pollution in China: Evaluation, Convergence Characteristics and Spatial Effects . Chinese Geographical Science, 2021, 31(3): 571-584. doi: 10.1007/s11769-021-1200-1
    [14] Jiawei WANG, Shilin YE, Xinhua QI.  Regional Equity and Influencing Factor of Social Assistance in China . Chinese Geographical Science, 2021, 31(4): 611-628. doi: 10.1007/s11769-021-1195-7
    [15] SUN Zhe, ZHAN Dongsheng, JIN Fengjun.  Spatio-temporal Characteristics and Geographical Determinants of Air Quality in Cities at the Prefecture Level and Above in China . Chinese Geographical Science, 2019, 20(2): 316-324. doi: 10.1007/s11769-019-1031-5
    [16] DIAO Shuo, YUAN Jiadong, WU Yanyan.  Performance Evaluation of Urban Comprehensive Carrying Capacity of Harbin, Heilongjiang Province in China . Chinese Geographical Science, 2019, 20(4): 579-590. doi: 10.1007/s11769-019-1056-9
    [17] XUE Shuyan, LI Gang, YANG Lan, LIU Ling, NIE Qifan, Muhammad Sajid MEHMOOD.  Spatial Pattern and Influencing Factor Analysis of Attended Collection and Delivery Points in Changsha City, China . Chinese Geographical Science, 2019, 29(6): 1078-1094. doi: 10.1007/s11769-019-1086-3
    [18] LI Bo, SHI Zhaoyuan, TIAN Chuang.  Spatio-temporal Difference and Influencing Factors of Environmental Adaptability Measurement of Human-sea Economic System in Liaoning Coastal Area . Chinese Geographical Science, 2018, 28(2): 313-324. doi: 10.1007/s11769-018-0948-4
    [19] YU Chao, MA Yanji.  Carbon Emission Trends of Manufacturing and Influencing Factors in Jilin Province, China . Chinese Geographical Science, 2016, 26(5): 656-669. doi: 10.1007/s11769-016-0823-0
    [20] DAI Dandan, ZHOU Chunshan, YE Changdong.  Spatial-temporal Characteristics and Factors Influencing Commuting Activities of Middle-class Residents in Guangzhou City, China . Chinese Geographical Science, 2016, 26(3): 410-428. doi: 10.1007/s11769-016-0806-1
  • 加载中
图(3) / 表ll (6)
计量
  • 文章访问数:  291
  • HTML全文浏览量:  109
  • PDF下载量:  27
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-04-07
  • 录用日期:  2021-06-22
  • 刊出日期:  2021-09-05

Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China

doi: 10.1007/s11769-021-1227-3
    基金项目:  Under the auspices of Chinese Ministry of Education Humanities and Social Sciences Project (No. 19YJCZH241), Project of Chongqing Social Science Planning Project of China (No. 2020QNGL38), Science and Technology Research Program of Chongqing Education Commission of China (No. KJQN201901143), Humanities and Social Sciences Research Program of Chongqing Education Commission of China (No. 20SKGH169)
    通讯作者: ZHANG Fengtai. E-mail: zhfthero45@cqut.edu.cnGAO Lei. E-mail: lei.gao@csiro.au

English Abstract

MA Dalai, ZHANG Fengtai, GAO Lei, YANG Guangming, YANG Qing, AN Youzhi, 2021. Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China. Chinese Geographical Science, 31(5): 795−814 doi:  10.1007/s11769-021-1227-3
Citation: MA Dalai, ZHANG Fengtai, GAO Lei, YANG Guangming, YANG Qing, AN Youzhi, 2021. Spatiotemporal Dynamics of Green Total-factor Water-use Efficiency and Its Influencing Factors in China. Chinese Geographical Science, 31(5): 795−814 doi:  10.1007/s11769-021-1227-3
    • China is the most populous and largest developing country in the world. It has a significant demand for water resources because of its large population. Water consumption in China has been rapidly increasing owing to the prospering industry and agriculture, as well as improving living standards. From 2000 to 2018, China’s total water consumption increased by 9.4%, from 549.76 to 601.55 billion m3. China consumed 21.9% of its total water resources (2746.25 billion m³) in 2018, which exceeded the 20% threshold for a country to face water shortage (NBSC, 2019). Moreover, China’s per-capita water resources are lower than the global average and are unevenly distributed between regions (Mei et al., 2014). Therefore, water shortages have become a bottleneck for sustainable water use in China. It is imperative to address the challenge of water shortage by improving water-use efficiency.

      On the one hand, the utilization of water resources creates economic wealth, and on the other hand, generates pollutants such as wastewater, organic pollutants, and nitrogen oxides (Bryan et al., 2018). In 2017, China discharged 69.966 billion t of wastewater, exhibiting an increase of 68.53% from 41.516 billion t in 2000 (NBSC, 2001a–2018a). Various pollutants in water pose another challenge to water utilization. Despite including all possible inputs, the traditional evaluation methods for water-use efficiency only consider the economic value of water use (Hu et al., 2006), failing to address the environmental cost. This may lead to a bias in the evaluated efficiency (Li and Lin, 2015). To overcome this shortcoming, it is necessary to include both pollution output and economic output in the evaluation index system for green total-factor water-use efficiency (GTFWUE).

      A fundamental solution to ensure the sustainability of water resources is by improving water-use efficiency (Zoebl, 2006; Wang et al., 2014; 2019). The development of a suitable evaluation index system for water-use efficiency has been receiving increased attention. For example, some scholars respectively use single factor indicators such as water consumption per 10 000 yuan (RMB) of GDP and industrial water reuse rate to express industrial water use efficiency (Fujii et al., 2013; Zhao et al., 2016). These water-use efficiency indices are all single-factor indices. The single-factor evaluation method is simple in calculation, but overlooks the critical role of capital and manpower in water utilization. Compared with single-factor indices, total-factor water-use efficiency indices can reflect the effectiveness of water resources and many other factors in the production process (Hu et al., 2006). However, the measurement of the total-factor water-use efficiency is complex, and requires special methods. Wang et al. (2019) relied on a stochastic frontier analysis (SFA) to estimate the agricultural water use efficiency of China, while others employed data envelopment analysis (DEA) to measure water-use efficiencies in Kenya, Australia, and Tunisia (Chemak et al., 2010; Njiraini and Guthiga, 2013; Azad et al., 2015). Compared with SFA, DEA is an effective nonparametric statistical method to deal with multi input and multi output evaluation problems, so it has been more widely used in water resources efficiency evaluation (Wang et al., 2021).

      Existing studies have explored water-use evaluation in three key aspects: agricultural, industrial, and domestic water. The utilization efficiency of agricultural water has long been a research hotspot, as agriculture consumes the largest proportion of water resources, and agricultural water use is vital to food production (Gao and Bryan, 2017). In the field of agriculture, most scholars use DEA model to estimate agricultural water use efficiency (Wang G F et al., 2015; Cao et al., 2018; Geng et al., 2019; Zhang et al., 2019). In addition, a few studies focus on agricultural irrigation water efficiency (Yang and Jiang, 2016; Liang et al., 2018). With the progress of industrialization, industrial water consumption has been rapidly increasing, generating research interest in industrial water-use efficiency. One type of research analyzes industrial water use efficiency at the regional level by provincial data in China (Zhou et al., 2019; Liu et al., 2020; Qi and Song, 2020; Yang F et al., 2021). The other type of research analyzed water use efficiency in specific industrial sectors (Fujii et al., 2013; Kurle et al., 2015; Zhao et al., 2015). The increasing demand for domestic water by urban residents has highlighted the importance of improving the utilization efficiency of domestic water in urban areas. Shi et al. (2015) assessed the urban water resource utilization efficiency of 316 cities in China between 2000 and 2012, using DEA. Wibowo and Alfen (2015) constructed a two-stage Stackelberg DEA model to compare domestic water-use efficiencies in 269 Indonesian cities.

      Pollution is usually modeled as an undesirable output in efficiency evaluation. Zou and Cong (2020) adopted a directional distance function (DDF) model to measure industrial water resource utilization efficiency (IWRUE) of 30 provinces/municipalities/autonomous regions (hereafter provinces) in China (not including Tibet, Hong Kong, Macao, Taiwan as no data). Nevertheless, the measured efficiency was biased because the slackness of inputs and desirable outputs were not considered in the function, which was a radial and angular measuring method. In addition, some scholars adopted the linear data conversion method to convert sewage discharges into desirable output (Ma et al., 2016; Zhang X Y et al., 2020). However, the above method for the undesirable output has some advantages, but it still deviate from the production law, so the efficiency value also has some deviation (Zeng and Wei, 2021). To effectively solve this bias, Tone (2001) proposed a non-radial, non-angular slack-based measure (SBM). In the SBM, the undesirable output was included as an output index in the efficiency calculation, and the slackness of inputs and outputs were also considered. After this, SBM is the most common model for measuring water-use efficiency, which contains undesirable outputs (Sun et al., 2014a; Wang Y S et al., 2015; Deng et al., 2016). The SBM evaluates the relative efficiency of each decision-making unit (DMU) according to the maximum distance of all projection points of multiple DMUs to the optimal frontier. Therefore, before applying SBM, a set of multiple DMUs must be established. Notably, the SBM might underestimate the efficiency of DMUs (Wang et al., 2013). Once the efficiency is underestimated, the input and output variables must be significantly improved to optimize the efficiency, which incurs additional economic costs (Liu et al., 2020). To address this issue, this study selected a minimum distance to the strong frontier model (referred to as mSBM) to measure the GTFWUE with wastewater discharge as the undesirable output.

      In terms of the influencing factors of water-use efficiency, some scholars investigated how the efficiency is impacted by marketization (Chen et al., 2019), high-quality urbanization (Zhang X L et al., 2020), and water rights trading policy (Chen et al., 2021). Some scholars had analyzed two main factors that affect water use efficiency (Ding et al., 2019; Wei et al., 2019; Yang et al., 2019). In addition, some scholars discussed composite factors that influence water use efficiency (Sahin et al., 2015; Zhao et al., 2017a; Song et al., 2018; Xu and Liang, 2020). Most of these studies adopt non-spatial measurement methods, such as Tobit model and ordinary least squares (OLS) model, to perform empirical analysis of the influencing factors. These non-spatial measurement methods do not consider the possible spatial correlation between observation values, which might lead to certain variations in the estimation results of the model (Elhorst, 2010).

      In view of this, this study was based on the findings of Yao et al. (2018), who was one of the earliest scholars to propose the concept of green total-factor water efficiency (GTFWE). A GTFWUE evaluation index system with waste water as an undesirable output is further proposed. Furthermore, this study uses the mSBM model to measure GTFWUEs of 30 provinces in China from 2000 to 2018, and analyzes their spatial correlation. Finally, a spatial econometric model is constructed to study the factors influencing the GTFWUE. Through this study, on the one hand, it can correct the problem that the efficiency value is too small when most previous studies used SBM method to evaluate water use efficiency, so as to reduce the change range of input-output variables in the process of efficiency improvement. On the other hand, this study also helps to solve the drawbacks of the non-spatial measurement method, that is, not considering the possible spatial correlation of observed variables. This can more accurately analyze the influencing factors of water use efficiency. The study could provide valuable enlightenment for the formulation of China’s future water-saving policy, and has certain guiding significance for the realization of water resources sustainable utilization.

    • The research area of this study is China’s 30 provincial-level administrative regions (hereafter provinces), owing to the lack of relevant data for Tibet, Hong Kong, Macao, and Taiwan, these areas are not taken as the sample area in the study (Fig. 1). In view of the differences in economic development level, administrative division and geographical location, the whole country is divided into the eastern region (Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan), the central region (Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan), and the western region (Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang) (Zhao et al., 2019; Yang J M et al., 2021). In order to study the regional differences of GTFWUEs from 2000 to 2018, this study also made a comparative analysis of GTFWUEs in the eastern, central, and western regions.

      Figure 1.  Location and geographical divisions of China. Tibet, Hong Kong, Macao and Taiwan of China are not included as no data

    • The statistical data in this study mainly comes from China Statistical Yearbook (NBSC, 2001a–2019a), China Statistical Yearbook on Science and Technology (NBSC, 2001b–2019b), China Population and Employment Statistics Yearbook (NBSC, 2001c–2019c), China Yearbook on Environment (NBSC and MEE, 2001–2019), and statistics of various provinces, municipalities and autonomous regions in China. Missing data were interpolated using the data of adjacent years.

    • According to Hu et al. (2006), GTFWUE is also known as the green total factor productivity (TFP) of water use. Specifically, the study defines GTFWUE as the ratio of optimal water consumption to the actual water consumption when economic output is optimized and environmental pollutants are minimized so that labor, capital, and other input factors remain unchanged. In summary, the GTFWUE reflects the comprehensive water utilization efficiency under the constraints of optimal economic output and minimal pollutant emissions. The measurement accuracy of GTFWUE relies on a sound evaluation index system. Our evaluation index system consists of two parts: input and output variables (Yao et al., 2018; Yang F et al., 2021). The input variables include the labor force, capital stock, and water consumption, while the output variables include the desirable output (economic value) and an undesirable output (wastewater discharge) (Table 1).

      Table 1.  Statistical description of the input and output variables for measuring green total-factor water-use efficiency (GTFWUE) in China

      VariableUnitObs.MinMaxMeanSD
      Labor force 104 person 570 275.50 6766.00 2530.04 1680.30
      Capital stock 108 yuan (RMB) 570 211.80 192708.40 30437.67 36245.90
      Water consumption 108 m3 570 19.96 592.00 194.04 136.59
      Economic value 108 yuan (RMB) 570 263.59 58545.69 10201.68 10373.25
      Wastewater discharge 108 t 570 1.10 93.83 19.68 15.98
      Notes: Obs. is the number of observation samples. SD is the standard deviation

      (1) Labor force: The labor force is characterized by the number of employees at the end of the year in each province.

      (2) Capital stock: The capital stock was estimated using the permanent inventory method (PIM):

      $$ {K_{i,t}} = {I_{i,t}} + (1 - \delta){K_{i,t - 1}} $$ (1)

      where Ki,t and Ii,t are the capital stock and investment of province i in period t, respectively; Ki,t−1 is the capital stock of province i in period t−1; δ is the capital depreciation rate of province i in the t-th period, according to Shan (2008), its value is 10.96% in this study. To eliminate inflation caused by price factors, the nominal capital stock was converted into actual capital stock with 2000 as the base period.

      (3) Water consumption: Water consumption is the sum of agricultural, industrial, domestic, and ecological water consumption.

      (4) Economic value: The economic value is characterized by the gross domestic product (GDP) of each province. To eliminate the inflation caused by price factors, the nominal GDP of each province was converted into actual GDP, with 2000 as the base period.

      (5) Wastewater discharge: Wastewater discharge is an undesirable, which refers to the wastewater discharged by each province in water consumption, including industrial wastewater, domestic wastewater, organic pollutants, and ammonia nitrogen.

    • Considering the significant relationships among water utilization and economy, technology, society, and government policies (Xie et al., 2019), the potential factors influencing the GTFWUE investigated in this study were: economic growth, population size, water endowment, water-use structure, technological progress, opening-up level, government influence, and urbanization level. All the indications are shown in Table 2. Each influencing factor is explained as follows.

      Table 2.  Statistical description of factors influencing on green total-factor water-use efficiency (GTFWUE) in China

      VariablesUnitObs.MinMaxMeanSD
      Economic Growth (EG) yuan (RMB) 570 7.92 11.85 10.08 0.85
      Population Size (PS) 104 person 570 6.25 9.34 8.16 0.76
      Water Endowment (WE) m3/person 570 5.08 7.89 6.04 0.57
      Water-use Structure (WS) % 570 10.69 95.19 61.51 17.39
      Technological Progress (TP) 108 yuan (RMB) 570 1.95 13.08 8.72 1.88
      Opening-up Level (OL) % 570 0.04 15.36 2.55 2.23
      Government Influence (GI) % 570 0.42 18.97 7.77 4.94
      Urbanization Level (UL) % 570 23.20 89.60 49.96 15.13

      (1) Economic growth (EG): An increase in economic growth would lead to an improvement in the water-use efficiency and water resource allocation (Dong et al., 2014). EG elevates the income and enhances the environmental awareness of residents, leading to an improvement in the water-use efficiency by adopting resource-conserving and environment-friendly production methods. Herein, per-capita GDP is selected as the proxy variable for EG. To eliminate collinearity, the natural logarithm of per-capita GDP was adopted in our empirical model. The coefficient of EG was assumed to be positive.

      (2) Population size (PS): The greater the PS, the smaller the per-capita water resources required, and the higher the regional water price. A high water price tends to promote the TFP of water resources (Molinos-Senante et al., 2014). In this study, the year-end permanent population of each province was selected as the substitute variable of PS and used in the natural logarithmic form for our empirical model. The coefficient of PS was assumed to be positive.

      (3) Water endowment (WE): The study has shown that water-rich regions in China tend to be inefficient in water use because the local people have poor water conservation awareness, and the situation is exactly the opposite in water-deficient regions (Ding et al., 2018). Herein, the per-capita water resource of each province was chosen as the substitute variable of WE and used in the natural logarithmic form for our empirical model. The coefficient of WE was assumed to be negative.

      (4) Water-use structure (WS): Agricultural water accounts for the largest proportion and is a defining feature of the regional WS. Over 84% of agricultural water is consumed by farmland irrigation. However, water is supplied to most farmlands in China through flood irrigation, which is inefficient and wasteful. Therefore, China has great potential for conserving agricultural water (Kaneko et al., 2004). The coefficient of WS was assumed to be negative.

      (5) Technological progress (TP): TP stimulates innovation of production technology and upgrading of high-consumption equipment, which in turn promotes production efficiency and reduces water consumption in production (Jefferson et al., 2006). In the study, the TP in each province was measured by the natural logarithm of the research and development (R&D) expenditure. The coefficient of TP was assumed to be positive.

      (6) Opening-up level (OL): The level of opening-up a region is largely reflected in the introduction of foreign investment. As China undergoes further opening-up, the substantial inflow of foreign capital is accompanied by the absorption of advanced production technology, mature management practices, and strict environmental standards. These are conducive to resource conservation and pollution control (Mielnik and Goldemberg, 2002). OL in each province was characterized by ratio of the actual FDI to GDP. Here, actual FDI converts from USD to RMB at the mean interest rate. The coefficient of OL was assumed to be positive.

      (7) Government influence (GI): To ensure the sustainability of water utilization, it is necessary to improve the management efficiency of water resources by government intervention (Deason et al., 2001). The government should avoid excessive intervention in the market, which might distort the efficiency of water resource allocation. Here, the GI in each province was characterized by the expenditure on agricultural and forestry water as a proportion of the general budget. It was assumed that the coefficient of GI was variable.

      (8) Urbanization level (UL): In China, urban residents are generally better educated than their rural counterparts. Therefore, urban residents have a relatively high awareness of water conservation and seldom waste domestic water. The higher the UL, the better is the water-use efficiency. UL in each province was substituted by the permanent urban population as a proportion of the total population. The coefficient of UL was assumed to be positive.

    • The minimum distance to the strong frontier model (mSBM) is an effective improvement of the maximum distance to the frontier model. The distance to the strong frontier model selects the projection point with the minimum distance from the front surface as reference; thus, the efficiency measured by the model is relatively high. This implies that the invalid DMU can be improved to a DMU only by making small changes to the input-output variables. This greatly reduces the cost of the change.

      Based on the study by Jahanshahloo et al. (2012), this study selected a mSBM with an undesirable output. The mSBM determines the projection point closest to the optimal frontier among numerous DMU projection points in the production set, and then measures the efficiency of the DMUs (Yang G M et al., 2021). We assume that there is a production set of n DMUs, each of which generates s1 unit of the desirable output and s2 unit of the undesirable output from m units of input. For convenience, the inputs, desirable output, and undesirable output are expressed as vectors, that is $X = ({x_1},{x_2},...,{x_n}) \in R_ + ^{m \times n}$, ${Y^g} = (y_1^g,y_2^g,...,y_n^g) \in R_ + ^{{s_1} \times n}$, and ${Y^b} = (y_1^b,y_2^b,...,y_n^b) \in R_ + ^{{s_2} \times n}$, respectively. If ${\rm{DM}}{{\rm{U}}_0} = ({x_0},y_0^g,y_0^b)$ is the final set of DMUs, ${{{P}}^t}(x) = \left\{ {(x,{y^g},{y^b}):x\; {\rm{can}}\; {\rm{product}}\; {y^g}\; {\rm{and}}\; {y^b}} \right\}$ is the set of all possible production scenarios, Fs(P) is the set of all outputs in the production set lying on the optimal frontier, and L1 is the shortest distance from the projection points of the DMUs to the optimal frontier, the mSBM can be expressed as follows:

      Objective function:

      $$ \begin{split} ({\rm{mSBM}})& \min \left(\displaystyle\sum\limits_{i = 1}^m {s_{i0}^ - } + \displaystyle\sum\limits_{r = 1}^{{s_1}} {s_{r0}^ + } + \displaystyle\sum\limits_{l = 1}^{{s_2}} {s_{l0}^ - }\right) + \\ &{\rm{M}}\left(\displaystyle\sum\limits_{i = 1}^m {\overline s _{i0}^ - } +\displaystyle\sum\limits_{r = 1}^{{s_1}} {\overline s _{r0}^ + } + \displaystyle\sum\limits_{l = 1}^{{s_2}} {\overline s _{l0}^ - }\right) \\ &s_{i0}^ - \ge 0,\; \; \; i = 1,...,m \\ & s_{r0}^ + \ge 0,\; \; \; r = 1,...,{s_1} \\ & s_{l0}^ - \ge 0,\; \; \; l = 1,...,{s_2} \\ \end{split} $$ (2)

      Constraints:

      $$\begin{array}{l} \; \; \; \; \max \left(\displaystyle\sum\limits_{i = 1}^m {\overline s _{i0}^ - } + \displaystyle\sum\limits_{r = 1}^{{s_1}} {\overline s _{r0}^ + } + \displaystyle\sum\limits_{l = 1}^{{s_2}} {\overline s _{l0}^ - }\right) \\ \; \; \; \; \; \; {s_ \cdot }{t_.}\; \displaystyle\sum\limits_{j \in {E_c}} {{\lambda _j}{x_{ij}}} + \overline s _{i0}^ - = {x_{i0}} - s_{i0}^ - \\ \; \; \; \; \; \; \displaystyle\sum\limits_{j \in {E_c}} {{\lambda _j}y_{ij}^g - } \overline s _{r0}^ + = y_{i0}^g + \; \; s_{r0}^ + \\ \; \; \; \; \; \; \; \displaystyle\sum\limits_{j \in {E_c}} {{\lambda _j}y_{ij}^b + } \overline s _{l0}^ - = \; y_{i0}^b\; - s_{l0}^ - \\ \; \; \; \; \; \; \; {\lambda _j} \ge 0,\; \overline s _{i0}^ - \ge 0,\; \; \overline { s} _{r0}^ + \ge 0,\; \overline {s} _{l0}^ - \ge 0 \\ \end{array} $$ (3)

      where $s_{i0}^ - $, $s_{r0}^ + $, $s_{l0}^ - $, $\overline s _{i0}^ - $, $\overline s _{r0}^ + $ and $\overline s _{l0}^ - $ are the slack terms of the input and output variables, respectively, and M is the constant term, which is a large positive number. i, r and l represent i-th input element, r-th desirable output and l-th undesirable output respectively. xij, $y_{ij}^g$ and $y_{ij}^b$ represent the input elements, the desirable outputs and the undesirable outputs of the j-th DMU respectivel. ${x_{i0}}$, $y_{i0}^g$ and $y_{i0}^b$ are the input elements, the desirable outputs and the undesirable outputs of final DMU respectively. λj is the weight of the j-th DMU, and Ec is the set of all DMUs. Equations (2) and (3) form a typical two-layer linear programming structure.

      mSBM is closely related to the SBM. The former is based on the minimum distance to a strong efficient frontier, while the latter is based on the maximum distance to a strongly efficient frontier. The two models differ in the selection of projection points. The mSBM selects the minimum distance L1, whereas the SBM chooses the maximum distance L2. As an improved version of the SBM, the mSBM optimizes the efficiency of DMUs with minimal changes to the inputs or outputs, thereby changing inefficient DMUs to efficient ones. Hence, mSBM can significantly reduce the economic cost of the producer.

    • (1) Global Moran’s I

      The objective of this study was to verify the existence of significant spatial correlations among the GTFWUEs of different regions. Spatial correlation is also called spatial dependence, which refers to the potential interdependence of the observation data of some variables within the same distribution area. Due to the impact of spatial interaction and diffusion, the observation data of these variables reflected spatially may not be independent, but related to each other. Anselin and Getis (1992) also emphasized that most spatial data have strong or weak spatial dependence. In this study, we adopted Global Moran’s I to measure the spatial correlation of GTFWUE (Moran, 1948):

      $$I = \dfrac{n}{{\displaystyle\sum\limits_{i = 1}^n {{{({x_i} - \overline x)}^2}} }}\dfrac{{\displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {{W_{ij}}({x_i} - \overline x)({x_j} - \overline x)} } }}{{\displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {{W_{ij}}} } }}$$ (4)

      where n is the number of all provincial administrative regions (provinces) in this study; xi and xj are the observed values of provinces i and j, respectively; $\overline x = \left(\displaystyle\sum\nolimits_i {{x_i}}\right)/n$ is the mean observed value of all provinces. The value of Global Moran’s I lies within the interval [–1, 1]. If the index value is –1, the observed values of the provinces are negatively correlated with each other spatially; if the index value is 1, the observed values of the provinces are positively correlated with each other spatially; and if the index value is 0, the observed values of the provinces are independent of each other, exhibiting no spatial correlation. Wij is a spatial weight matrix. There are three forms of the spatial weight matrix: distance, economic, and adjacency matrices. The latter is the most widely used and easiest to calculate. Therefore, the spatial adjacency matrix of the ones and zeros was selected as follows:

      $$ {W_{ij}} = \left\{ \begin{array}{l} 1\; \; \Pr {\rm{ovince}}\; i\; {\rm{is}}\; {\rm{adjacent}}\; \; {\rm{to}}\; {\rm{province}}\; j\; \\ 0\; \; \Pr {\rm{ovince}}\; i\; {\rm{is}}\; {\rm{not}}\; {\rm{adjacent}}\; \; {\rm{to}}\; {\rm{province}}\; j\; \\ \end{array} \right. $$ (5)

      For the significance test, the expectation and variance of Global Moran’s I should be calculated by Equations (6) and (7), respectively.

      $$ {{E}}(I) = - \frac{1}{{n - 1}} $$ (6)
      $$ {\rm{Var(}}I{\rm{) = }}\frac{{{n^2}{w_1} + {n^2}{w_2} + 3w_0^2}}{{w_0^2({n^2} - 1)}} - {{{E}}^2}(I) $$ (7)

      where n represents the number of provinces to be investigated. ${w_0} = \displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {{w_{ij}}} } $, ${w_1} = \dfrac{1}{2}{\displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {({w_{ij}} + {w_{ji}})} ^2} }$; ${w}_{2}= $$ \displaystyle \sum _{i=1}^{n}\displaystyle \sum _{j=1}^{n} ({w}_{i{\rm{r}}}+{w}_{i{\rm{c}}})^{2}$. wij is the weight of i-th row and j-th column in the spatial weight matrix, and wji is the weight of j-th row and i-th column in the spatial weight matrix. ${w}_{i{\rm{r}}}$ and ${w}_{i{\rm{c}}}$ are the sums of the numbers in the i-th row and i-th column of all spatial matrices, respectively.

      The values of Global Moran’s I must qualify the significance test to verify the authenticity of the index. The most common method to test the significance of Global Moran’s I is the Z-score normalization.

      $$ {{Z}}(I) = \frac{{\left[ {I - {{E}}(I)} \right]}}{{\sqrt {{\rm{Var(}}I{\rm{)}}} }} = \frac{{\left[ {I - {{E}}(I)} \right]}}{{{{SD}}(I)}} $$ (8)

      where E(I) is the expectation of the Global Moran I. ${{SD}}(I) = \sqrt {{\rm{Var(}}I{\rm{)}}} $, wihch represents the standard deviation of the Global Moran’ I. The Global Moran’s I value is authentic if its Z-score passes the tests at the 1%, 5%, and 10% significance levels. In this case, the objects investigated had significant spatial correlations.

      (2) Local indication of spatial association (LISA) map based on Local Moran’s I

      Global Moran’s I only reflects the spatial correlations between the observed values in different provinces on a global scale. It can not characterize spatial correlations at the local scale. To overcome this drawback, the Local Moran’s I of the observed value of each province was calculated and plotted as a LISA map to determine the spatial distribution of the observed values. The LISA map contained four regions types: high-high (H-H) cluster region, low-high (L-H) cluster region, low-low (L-L) cluster region, and high-low (H-L) cluster region. Each province in the H-H/L-L cluster region as a relatively high/low observed value, and its neighboring provinces also have relatively high/low observed values. Each province in the H-L/L-H cluster region has a relatively high/low observed value, but its neighboring provinces have relatively low/high observed values. Local Moran’s Ii of province i can be calculated according to the following equation (Moran, 1950):

      $$ {I_i} = {Z_i}\sum\limits_{j = 1}^m {{w_{ij}}} {Z_j} $$ (9)

      where Zi and Zj are the normalized observed values of provinces i and j, respectively; $m$ is the number of all provinces investigated and wij is the normalized row of the spatial weight matrix. Province i belongs to the H-H cluster region, if Local Moran’s Ii > 0 and Zi > 0, the L-L cluster region if Local Moran’s Ii > 0 and Zi < 0, the H-L cluster region if Local Moran’s Ii < 0 and Zi > 0, and the L-H cluster region if Local Moran’s Ii < 0 and Zi < 0.

      (3) Spatial econometrics model

      To effectively improve China’s GTFWUE, the key is to understand the factors affecting GTFWUE. To this end, it is necessary to choose a suitable model to verify the significance of the influencing factors. Here, the ordinary panel econometric model (Eq. (10)) is adopted to evaluate the factors affecting China’s GTFWUE:

      $$ GTFWU{E_{it}} = {\beta _0} + {\beta _1}{X_{it}} + {\varepsilon _{it}} $$ (9)

      where X is the set of factors affecting the GTFWUE; β0 is the constant term; β1 is the coefficient term; ε is the random error term; and i and t are the subscripts of province and time, respectively.

      Traditionally, significance is tested through an OLS estimation. However, this method is not applicable to the model represented by Equation (10) if China’s GTFWUE is proven to have significant spatial correlation. This is because the OLS estimation only holds when the explained variables are spatially independent, that is, they are not spatially correlated. If the explained variables are spatially correlated, the OLS estimation might significantly differ from the actual result.

      Spatial econometric models can effectively solve this problem. The commonly used econometric models include spatial autoregressive and spatial error models. The spatial autoregressive model can elucidate the endogenous dependence of dependent variables, while the spatial error model can reveal the interaction effect of error terms (Zhou et al., 2017; Guo et al., 2021). The former mainly analyzes the behavior of adjacent regions and its impact on the behavior of other regions in the entire system; the latter mainly studies the relationship between regions through error terms. The basic form of the spatial autoregressive model is as follows (Anselin, 1988):

      $$\left\{ \begin{array}{l} y = \rho Wy + \beta X + \varepsilon \\ \varepsilon \sim N(0,\; {\sigma ^2}{I_n}) \\ \end{array} \right.$$ (11)

      where y is the explained variable, ρ is the spatial autoregressive model parameters reflecting the significance of the spatial effect, and W is an n × n spatial weight matrix. X is a set of explanatory variables, and β are coefficients of explanatory variables. N is normal distribution. σ is the standard deviation, and I is the identity matrix. n is the number of samples to be investigated. Here, W is expressed as a spatial adjacency matrix of zeros and ones.

      The basic form of the spatial error model is as follows (Haining, 1993):

      $$ \left\{ \begin{array}{l} y = \beta X + u \\ u = \lambda Wu + \varepsilon \\ \varepsilon \sim N(0,\; {\sigma ^2}{I_n}) \\ \end{array} \right. $$ (12)

      where λ is the spatial error coefficient, u is a random interference term in accordance with the normal distribution. The other variables are same as defined for Equation (11). The result of λ in the significance test indicates whether the model has a spatial effect.

    • The mSBM model was adopted to calculate the GTFWUE of 30 provinces of China from 2000 to 2018. The GTFWUEs were found to have significant provincial differences (Table 3). During the study period, the mean GTFWUEs of four provinces (Tianjin, Liaoning, Shanghai, and Yunnan) were 1, which was the optimal frontier. However, those of other provinces failed to reach the frontier value, which showed that further improvement was required. In addition, the other four provinces (Fujian, Shandong, Zhejiang, and Heilongjiang) had relatively high GTFWUEs, averaging between 0.8 and 0.9. Most of the top eight provinces belonged to the eastern coastal area. Medium mean GTFWUEs, ranging 0.6–0.8, were observed for 16 provinces: Anhui, Sichuan, Hebei, Hubei, Guangdong, Chongqing, Jilin, Beijing, Hainan, Hunan, Jiangsu, Shanxi, Henan, Qinghai, Shaanxi, and Guizhou. These provinces are generally located in inland areas. Their mean GTFWUEs had considerable distances to the optimal frontier, which could be improved through improvement measures. Finally, six provinces located in underdeveloped inland areas, namely, Inner Mongolia, Guangxi, Jiangxi, Xinjiang, Gansu, and Ningxia, had the lowest mean GTFWUE not exceeding 0.6.

      Table 3.  Green total-factor water-use efficiencies (GTFWUEs) of Chinese provinces during 2000–2018

      Province2000200220042006200820102012201420162018Average
      Tianjin 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
      Liaoning 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
      Shanghai 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
      Yunnan 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
      Fujian 1.0000 1.0000 0.7680 0.8872 0.8589 0.8646 0.7950 0.8108 0.8230 0.8582 0.8605
      Shandong 1.0000 1.0000 0.7779 0.7328 0.7848 0.7824 0.7406 0.7777 0.7924 0.8284 0.8263
      Zhejiang 0.9795 0.8774 0.7741 0.8357 0.8162 0.7860 0.7635 0.7942 0.8090 0.7706 0.8150
      Heilongjiang 0.8095 0.8120 0.7770 1.0000 1.0000 1.0000 0.6315 0.6186 0.6188 0.6232 0.8064
      Anhui 0.8033 0.8474 0.8867 0.8794 0.8658 0.7624 0.7803 0.8085 0.7096 0.6959 0.7996
      Sichuan 0.7429 0.7570 0.7647 0.7932 0.6878 0.7393 0.7869 0.8151 0.8262 0.8419 0.7745
      Hebei 0.9061 0.8659 0.8459 0.8609 0.7609 0.7436 0.6662 0.6861 0.6709 0.6367 0.7647
      Hubei 0.8162 0.7975 0.7712 0.8339 0.7578 0.7742 0.6883 0.7025 0.6880 0.6993 0.7575
      Guangdong 0.8562 0.7259 0.6689 0.7927 0.7767 0.7657 0.6513 0.6556 0.6580 0.6678 0.7179
      Chongqing 0.6436 0.6871 0.7056 0.7237 0.6530 0.7135 0.7356 0.7773 0.7249 0.7099 0.7119
      Jilin 0.8323 0.7956 0.7388 0.8088 0.7589 0.7568 0.5859 0.5979 0.5568 0.6219 0.7097
      Beijing 1.0000 0.7479 0.6607 0.6638 0.6740 0.6608 0.6769 0.6811 0.6376 0.7014 0.7089
      Hainan 0.7030 0.7572 0.6536 0.7906 0.7730 0.7239 0.6432 0.6507 0.6479 0.6663 0.7062
      Hunan 0.7553 0.7502 0.6926 0.7860 0.7414 0.6948 0.6386 0.6473 0.6678 0.6507 0.7055
      Jiangsu 0.9013 0.6978 0.6156 0.7424 0.7422 0.7538 0.6210 0.6347 0.6352 0.6489 0.6952
      Shanxi 0.7561 0.7550 0.7187 0.7209 0.6628 0.6283 0.5958 0.5985 0.5762 0.5969 0.6601
      Henan 0.8133 0.7603 0.7206 0.7497 0.6567 0.6240 0.5666 0.5809 0.5633 0.5803 0.6563
      Qinghai 0.6017 0.7136 0.6638 0.7222 0.6798 0.6494 0.5837 0.5830 0.5622 0.5722 0.6360
      Shaanxi 0.7970 0.7434 0.6321 0.6366 0.5642 0.5740 0.6000 0.6113 0.5853 0.5751 0.6286
      Guizhou 0.6461 0.6690 0.5771 0.6078 0.5856 0.6300 0.6398 0.6500 0.6010 0.6808 0.6260
      Inner Mongolia 0.7380 0.6567 0.5759 0.4898 0.6711 0.6936 0.4558 0.4553 0.4320 0.4870 0.5706
      Guangxi 0.6029 0.5889 0.5714 0.6252 0.6136 0.5711 0.4787 0.4989 0.5606 0.5492 0.5690
      Jiangxi 0.7373 0.5737 0.5182 0.5975 0.5913 0.5496 0.4954 0.4861 0.5083 0.5504 0.5560
      Xinjiang 0.6143 0.5940 0.5326 0.5434 0.5266 0.5228 0.4672 0.4577 0.4306 0.4299 0.5149
      Gansu 0.6125 0.5105 0.5358 0.5721 0.5770 0.5599 0.4263 0.4516 0.4323 0.4291 0.5117
      Ningxia 0.5170 0.4951 0.4647 0.4851 0.4684 0.4766 0.3758 0.3882 0.4045 0.4013 0.4524
      Note: Due to space limitations, only the green total-factor water-use efficiency values for the even-numbered years are listed. Not including Tibet, Hong Kong, Macao and Taiwan of China

      In summary, the provinces with high and low GTFWUEs were concentrated in the coastal and inland areas, respectively. The only exception was Yunnan, which belongs to the western inland areas, despite reaching the optimal frontier of mean GTFWUE. A possible reason for this exception is that Yunnan has high precipitation, consumes limited agricultural water, and has a low reliance on industry.

      The temporal GTFWUE of China and its three major regions (eastern, central, and western regions) are shown in Fig. 2. It can be observed that the overall nationwide GTFWUE trend was consistent with the regional trends. From 2000 to 2011, the GTFWUE of China and its three regions decreased slowly, except in 2005; from 2012 to 2016, the GTFWUE was generally stable; then, the GTFWUE rapidly increased in 2017 then decreased in 2018. Notably, China’s GTFWUE increased suddenly in 2005 because 2005 was the last year of China’s tenth Five-Year Plan, and the government paid greater attention to water conservation and solving water resource problems, by incorporating several major arrangements and maintaining a high level of investment in water conservation.

      Figure 2.  Green total-factor water-use efficiencies of different regions of China from 2000 to 2018

      Moreover, there were significant differences between the three major regions in the GTFWUE. During the study period, the mean GTFWUE of the eastern region was highest with a value of 0.8268. The value of the central region (0.7064) was close to the national average (0.7247), while that of the western region (0.6360) was considerably lower. The eastern region had a higher GTFWUE than the other two regions. From another perspective, the central and western regions had greater water-conservation potential than the eastern region. Hence, China should prioritize the central and western regions when developing water resource management policies.

    • The Global Moran’s I values of the GTFWUEs of China during the study period were obtained using GeoDa software (Fischer and Getis, 2009; Anselin et al., 2010). As shown in Table 4, the Global Moran’s I values were positive and significant at the 10%, 5%, and 1% levels throughout the study period, except in 2004 and 2011. Therefore, China’s GTFWUEs have significant positive spatial correlations and are spatially clustered with a certain regularity. Thus, the GTFWUE values of neighboring provinces are highly similar. These spatial correlations must be considered in the empirical analysis of the GTFWUE, resulting in more practically accurate empirical results.

      Table 4.  Global Moran’s I of green total-factor water-use efficiencies (GTFWUEs) in China

      YearMoran’s IE(I)MeanSD(I)Z(I)
      20000.4002–0.0345–0.04010.11913.6499***
      20010.2048–0.0345–0.03680.12321.9424***
      20020.1306–0.0345–0.03650.12451.3261*
      20030.1298–0.0345–0.03800.12621.3019*
      20040.0624–0.0345–0.04010.11770.8233
      20050.1043–0.0345–0.04270.12151.1424*
      20060.1538–0.0345–0.03790.11981.5718*
      20070.1876–0.0345–0.04020.12021.8478**
      20080.1176–0.0345–0.03830.12521.2149*
      20090.1899–0.0345–0.03860.11881.8889**
      20100.1021–0.0345–0.04650.12511.0919*
      20110.0780–0.0345–0.03720.12170.9244
      20120.1066–0.0345–0.03950.11931.1827*
      20130.0873–0.0345–0.04470.12181.0000*
      20140.1259–0.0345–0.03660.12061.3300*
      20150.1568–0.0345–0.03660.12111.5797*
      20160.1318–0.0345–0.03900.11971.3893*
      20170.1228–0.0345–0.04390.11921.3196*
      20180.1632–0.0345–0.04130.11561.7102**
      Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. The explanation of E(I), SD(I) and Z(I) can be found in Equations (6)–(8)
    • The Local Moran’s I values of China’s GTFWUE values in 2000 and 2018 were calculated and visualized using LISA maps (Fig. 3). LISA map of the provincial GTFWUEs in 2000 is presented in Fig. 3a. A total of 11 provinces (36.67%), namely, Beijing, Tianjin, Liaoning, Shandong, Fujian, Shanghai, Zhejiang, Hebei, Jiangsu, Jilin, and Henan, belonged in the H-H cluster region; five provinces (16.67%), namely, Anhui, Shanxi, Jiangxi, Hainan, and Guangxi, were in the L-H cluster region; ten provinces (33.33%), namely Shaanxi, Hunan, Sichuan, Inner Mongolia, Guizhou, Chongqing, Xinjiang, Qinghai, Gansu, and Ningxia, were in the L-L cluster region; and only four provinces (13.33%), namely, Yunnan, Guangdong, Hubei, and Heilongjiang, were in the H-L cluster region. Notably, most of the Chinese provinces (70.00%) were located in the H-H and L-L cluster regions. The provinces in the H-H cluster region are primarily concentrated in the eastern coastal areas, while the provinces in the L-L cluster region are mainly concentrated in the central and western regions, which are far from the economically developed coastal areas. Meanwhile, a few (30.00%) provinces belonged to the L-H and H-L cluster regions and had an insignificant clustering effect.

      Figure 3.  Local indication of spatial association maps of provincial green total-factor water-use efficiencies in China in 2000 and 2018

      The LISA map of the provincial GTFWUEs in 2018 is presented in Fig. 3b. A total of eight provinces (26.67%), namely, Beijing, Tianjin, Shanghai, Zhejiang, Anhui, Chongqing, Sichuan, and Yunnan, were in the H-H cluster region. Six provinces (20.00%), namely, Guizhou, Jiangsu, Hebei, Jilin, Jiangxi, and Guangxi, lay in the L-H cluster region. Twelve provinces (40.00%), namely, Guangdong, Hainan, Hunan, Henan, Heilongjiang, Shanxi, Qinghai, Shaanxi, Inner Mongolia, Gansu, Ningxia, and Xinjiang, belonged in the L-L cluster region. Only four provinces (13.33%), namely, Liaoning, Fujian, Shandong, and Hubei, were in the H-L cluster region. In summary, 66.67% of all the provinces belonged to H-H and L-L cluster regions in 2018, while only 33.33% belonged to the L-H and H-L cluster regions.

      Comparing these two LISA maps, it was found that the spatial distribution of some provinces (e.g., Liaoning, Shandong, Fujian, Hebei, Jiangsu, Jilin, Henan, Guangdong, and Heilongjiang) varied from 2000 to 2018. Compared with 2000, the number of provinces in the H-H cluster region decreased in 2018. Liaoning, Fujian, and Shandong shifted from the H-H cluster region to the H-L cluster region; Jiangsu and Jilin from the H-H cluster region to the L-H cluster region; and Henan from the H-H cluster region to the L-L cluster region. Moreover, Shanxi and Hainan shifted from the L-H cluster region to the L-L cluster region, and Guangdong and Heilongjiang changed from the H-L cluster region to the L-L cluster region. In summary, the spatial distribution of GTFWUE varied temporally, but approximately 2/3 of all provinces belonged to the H-H and L-L cluster regions in the two LISA maps, with only 1/3 belonging to the L-H and H-L cluster regions. Therefore, China’s provincial GTFWUEs exhibited both a significant clustering effect in local areas and a significant spatial heterogeneity.

    • The estimation and spatial autocorrelation test of Equations (10) were obtained using Matlab 7.12. An initial simulation was performed to verify the presence of spatial correlations between the residual terms. The simulation results are presented in Table 5. To reveal the necessity of controlling fixed effects, the estimation results of a non fixed effect model, a space fixed effect model, a time fixed effect model, and a two-way fixed effect model are also provided in Table 5. The results of the four fixed-effects models were compared to determine the model with the best fit.

      Table 5.  Estimation results of common panel data models

      VariablesNon fixed effectSpace fixed effectTime fixed effectTwo-way fixed effect
      Economic Growth (EG) 0.0205
      (1.1245)
      −0.0570***
      (−3.8886)
      0.0643**
      (2.3651)
      0.0246
      (0.9509)
      Population Size (PS) 0.1281***
      (11.3144)
      0.0240
      (0.3578)
      0.1313***
      (12.1319)
      0.0490
      (0.6928)
      Water Endowment (WE) −0.0536***
      (−5.3478)
      0.0328
      (0.8948)
      −0.0484***
      (−5.0122)
      0.0076
      (0.2285)
      Water-use Structure(WS) −0.0198
      (−0.4411)
      0.0170
      (0.2382)
      −0.0071
      (−0.1581)
      0.0539
      (0.0539)
      Technological Progress (TP) −0.0632***
      (−9.9968)
      −0.0020
      (−0.3104)
      −0.0708***
      (−11.4664)
      −0.0055
      (−0.7974)
      Opening-up Level (OL) 2.0881***
      (7.7781)
      0.2567
      (1.1345)
      1.8937***
      (6.7982)
      0.2301
      (1.1567)
      Government Influence (GI) −0.5671***
      (−3.5262)
      −0.2297**
      (−2.2727)
      −1.3898***
      (−4.9916)
      −0.7290***
      (−4.3731)
      Urbanization Level (UL) 0.6466***
      (6.9222)
      0.2701***
      (2.6005)
      0.5009***
      (5.0394)
      0.2384***
      (2.5729)
      R2 0.4878 0.2360 0.5096 0.0513
      Log-L 421.7723 787.1394 455.7593 867.5175
      DW 2.0404 1.3724 2.1387 1.8060
      LM-lag 0.0003 139.9561*** 5.6365** 27.2893***
      Robust LM-lag 3.1278* 2.4024* 0.9824 1.0044
      LM-err 0.9373 145.3999*** 4.7146** 29.2830***
      Robust LM-err 4.0648** 7.8461*** 0.0605 2.9981*
      Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. The Robust LM-lag is the robustness test on Lagrange multiplier (LM) of spatial lag, while Robust LM-err is the robustness test on LM of spatial err. If the statistic value of the Robust LM-lag is greater than that of Robust LM-err, the spatial autoregressive model has better explanatory power. On the contrary, spatial error model has more advantages in variable interpretation

      First, the R2 values of the four models were compared. It can be seen that the R2 values of the non fixed effect, space fixed effect, time fixed effect, and two-way fixed effect models were 0.4878, 0.2360, 0.5096, and 0.0513, respectively. The time fixed effect model had the highest result, that is, the best fit. Next, the four models were compared using the Durbin-Watson (DW) statistic. The time fixed effect model was found to have a better DW (2.1387) than the other three models. Last, the Log-L value of time fixed effect model was 455.7593, which was smaller than that of the space fixed effect and two-way fixed effect models, but larger than that of the non fixed effect model. It showed that the parameters of the time fixed effect model have good likelihood. Based on the above comprehensive comparison, the time fixed effect model was chosen to analyze the factors affecting GTFWUE.

      Table 5 also lists the results of the spatial autocorrelation tests on the residual terms. Based on the Lagrange multiplier (LM) statistic, the strengths of the spatial autoregressive and spatial error models were compared. As shown in Table 5, the LM-lag of the time fixed effect model was 5.6365, at the 5% significance level and the LM-err of the model was 4.7146, which is also significant at the 5% level. The results demonstrated that the residual terms of common panel data models had significant spatial autocorrelation, indicating the necessity of applying the spatial econometric model. Because the LM-lag is slightly higher than LM-err, and the Robust LM-lag is also greater than Robust LM-err, so the spatial autoregressive model obtained better estimation results than the spatial error model.

    • Using Equations (11) and (12), the model represented by Equations (10) was modified into two spatial econometric models, namely, a spatial autoregressive model and a spatial error model. The estimation results of the two models are presented in Table 6. W × dep. var.is the spatial lag term of the spatial autoregressive model, while spat. aut.is the spatial error term of the spatial error mode. When the coefficient of W × dep. var. passes the significance level test, the adoption of the spatial autoregressive model is appropriate. Otherwise, the coefficient of this variable can not be tested at a significant level, which indicates that the adoption of the spatial autoregressive model is not appropriate. Similarly, whether the coefficient of spat. aut. passes the significance test, which verifies the suitability of the spatial error mode (Zou and Cong, 2020). It can be seen that the W × dep. var. of the spatial autoregressive model and spat. aut. of the spatial error model were –0.1499 and –0.1640, respectively; both passed the significance test at the 1% level. The results further confirmed the suitability of the spatial econometric model. The spatial econometric models had greater R2 and larger t-statistics compared with the common panel data models, so that the signs of the estimation coefficients were similar. This implied that spatial econometric models are better estimation tools than common panel data models. In addition, the spatial autoregressive model achieved greater Log-L than the spatial error model, indicating that it has better performance in terms of explanatory power than the spatial error model. Therefore, the estimation results of the spatial autoregressive model were selected for the discussion.

      Table 6.  Estimation results of time fixed effect spatial econometric models

      VariablesSpatial autoregressive modelSpatial error model
      Economic Growth (EG) 0.0749***
      (2.7931)
      0.0751***
      (2.8254)
      Population Size (PS) 0.1343***
      (12.6068)
      0.1358***
      (12.8389)
      Water Endowment (WE) −0.0520***
      (−5.3816)
      −0.0430***
      (−4.7091)
      Water-use Structure (WS) −0.0271
      (−0.6165)
      −0.0460
      (−1.0886)
      Technological Progress (TP) −0.0708***
      (−11.6075)
      −0.0714***
      (−11.6439)
      Opening-up Level (OL) 1.9371***
      (7.0073)
      1.7588***
      (6.4369)
      Government Influence (GI) −1.3302***
      (−4.8535)
      −1.2172***
      (−4.5878)
      Urbanization Level (UL) 0.4904***
      (5.0112)
      0.4865***
      (5.0866)
      W× dep. var. −0.1499***
      (−3.0767)
      spat. aut. −0.1640***
      (−2.7519)
      R2 0.5533 0.5439
      Log-L 459.0934 458.8562
      Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. W × dep. var.is the spatial lag term of the spatial autoregressive model, while spat. aut.is the spatial error term of the spatial error mode. Log-L is the log likelihood function

      Economic growth had a positive impact on the GTFWUE at the 1% significance level, indicating that an increase in per-capita GDP promoted GTFWUE. Thus, economic growth provides a fundamental driving force for improving water-use efficiency (Wolfe et al., 2009). This is particularly true in China, wherein economically developed areas maintain a highly intensive mode of water utilization, because they have a relatively reasonable industrial structure, efficient industrial water usage, and advanced agricultural water usage.

      Population size had a significant positive correlation with GTFWUE, as expected in this study. The continued expansion of the population has increased water consumption, raising concerns about the existing water resources, and exerting pressure on future water security (Sahin et al., 2015). Thus, an increasing number of residents have perceived the imminence of a water crisis and have incorporated water conservation into their daily lives.

      Water endowment had a negative impact on the GTFWUE at a significance level of 1%, indicating that the growth in per-capita water resources hinders the improvement of the GTFWUE. This result was consistent with the fact that water-rich areas have a low opportunistic cost of water use (Ma et al., 2016); excessive water consumption is common in these areas in industrial, agricultural, and domestic scenarios, which distorted the effective allocation of water resources.

      The estimation coefficient of water-use structure was negative, yet it did not pass the significance test. Hence, water-use structure did not play an evident role in the GTFWUE. This might be related to the regional imbalance in agricultural development. In economically developed eastern coastal areas, significant progress has been made in modern agriculture. The utilization of agricultural water has improved owing to the growing scale and mechanization of agricultural production. In many parts of these areas, sprinkler and trickle irrigation are installed to optimize the efficiency of agricultural water use. In contrast, agricultural production methods are relatively underdeveloped in the central and western inland areas. Most farmlands continue to be watered through flood irrigation, causing substantial agricultural water wastage.

      Contrary to our assumption, technological progress had a significant inhibitory effect on GTFWUE. This might be attributed to the trend in R&D activities in China. According to Acemoglu et al. (2012), R&D can lead to both clean and polluting technologies. If an enterprise is involved in polluting technology at the beginning, R&D activities will aggravate rather than curb pollutant discharge. In most Chinese enterprises, R&D emphasizes profitable polluting technology over environmentally-friendly clean technology. Thus, in such cases, R&D activities by enterprises would only generate more wastewater.

      As expected, opening-up level contributed to GTFWUE. As previously mentioned, water-use efficiency in a region can be promoted by introducing more foreign capital. In China, the provinces that attract more foreign capital often have a high GTFWUE. For example, in 2018, Shanghai, whose GTFWUE was on the optimal frontier, had a significantly higher FDI-GDP ratio (0.0350 vs. 0.0004) than Gansu, whose mean GTFWUE was only 0.4921.

      Government influence had a negative impact on GTFWUE. This implied that in most provinces, the government allocates higher funds to agricultural and forestry water, affecting the allocation to water resources. Excessive expenditure hinders the promoting effects of government policies on water utilization. For example, the expenditure on agricultural and forestry water accounted for a small proportion of the general budget of Tianjin, whose mean GTFWUE remained high, but accounted for a high proportion in that of Ningxia, whose mean GTFWUE was only 0.4524.

      Urbanization level had a significant positive correlation with GTFWUE. As previously mentioned, urban residents in China are generally more aware of the importance of water conservation than rural residents. The water-use efficiency in urban areas was better because of the complete water conservation infrastructure. Moreover, urban areas have better wastewater treatment facilities.

    • This study analyzed the spatiotemporal dynamics of China’s GTFWUE, and the research results showed that within the sampling period, the GTFWUEs of different provinces in China varied significantly. The GTFWUEs of most eastern coastal provinces were significantly higher than those of the inland provinces. Yao et al. (2018) reached the same conclusion, and observed significant differences in the GTFWEs of most provinces in China. In terms of time evolution, the change trends of GTFWUEs of provinces in the three major regions of China are basically the same within the study period, and they all exhibited stage-wise variation characteristics. Simultaneously, the variations in the GTFWUEs of the three major regions in terms of the average value of GTFWUE was evident, showing the following trend: eastern region > central region > western region. Yao et al. (2018) also analyzed the regional GTFWE by dividing China into four regions: the eastern, central, western, and northeast regions. During the study period, the average GTFWE of the eastern region was the highest, the average GTFWE of the central region ranked second, followed by the northeast region, and the average GTFWE of the western region was the lowest. The findings of this study are consistent with the research results of Yao et al. (2018), which confirmed their validity. In summary, the regional differences in GTFWUE were very significant. To ensure sustainable use of water resources in China, wherein the per-capita water resources are low, the government must formulate differentiated water-conservation policies.

      In addition, the Global Moran’s I value confirmed that the spatial distribution of China’s GTFWUE exhibited typical spatial correlation characteristics. Sun et al. (2014a) also verified this, showing that the spatial distribution of water resource utilization efficiency is not entirely random, but exhibits spatial clustering at the 1% significance level. Moreover, the LISA map indicated the evident spatial clustering effect of China’s local GTFWUE. However, the high-value clustering of China’s GTFWUE gradually weakened with time, but the low-value clustering was strengthened, indicating that China’s GTFWUE required further improvement measures to be implemented.

      Most scholars investigating the factors influencing water utilization efficiency in China, have not included the spatial effect in their hypothesis. They used the OLS method that ignores the spatial effect in estimating the model, resulting in the problem of setting deviation in the model in the actual application, and leading to incomplete, unscientific, and unreliable analysis results and inferences about the factors influencing China’s water utilization efficiency (Sun et al., 2014b). To overcome this drawback, this study adopted a spatial econometric model to explore the influence of GTFWUE considering eight factors: economic growth, population size, water endowment, water-use structure, technological progress, opening-up level, government influence, and urbanization level. The empirical results showed that economic growth was the core factor influencing GTFWUE, and the increase in per-capita GDP was conducive to the improvement of GTFWUE. Future studies must establish and continually develop the green GDP assessment system to improve the quality and efficiency of economic development. Population size had a significant positive effect on GTFWUE, which contradicted the findings of Sun et al. (2020) that population size has a significant hindering effect on the green efficiency of China’s water resources, but the result is consistent with the research conclusions of Song et al. (2018). A possible reason for this variation is that, as previously mentioned, population growth aggravated the shortage of water resources, driving people to adopt resource-conserving and environment-friendly production methods. Water endowment had a significant negative inhibitory effect on GTFWUE, which implied that an increase in per-capita water resources was not conducive to the improvement of the green efficiency of water resources, and this is consistent with the results of Qian and He (2011). This finding significantly contributes to the formulation of water-conservation and water-use policies by the government. Areas with rich water resources should pay further attention to the problem of wastage of water resources. The influence of water-use structure on the GTFWUE was not significant. This result contradicted our common knowledge, because as previously explained, agricultural water use is presently the main source of water consumption in China, and usage methods greatly vary regionally. Technological progress had a significant negative effect on GTFWUE, indicating that R&D investments could not effectively improve GTFWUE. Guo et al. (2021) reached a similar conclusion when studying the impact of R&D on SO2 emissions. This is because the limited R&D investment of enterprises is mainly focused on improving productivity, while investment in environmental protection and clean technology remains inadequate. Opening-up level had a positive effect on promoting GTFWUE. Zhao et al. (2017b) reached the same conclusion. In particular, as the Chinese government has put forward the concept of green development and gradually improved the FDI introduction standards and quality, the FDI innovation effect has gradually shifted to the ecological innovation spillover effect, which is conducive to environmental pollution reduction. Government influence had an adverse effect on GTFWUE, indicating that the government’s direct and substantial expenditure on agricultural and forestry water can not improve water utilization efficiency, and might lead to an adverse effect. Therefore, the government must adopt more appropriate measures, such as formulating preferential tax policies that encourage enterprises to implement water-conservation and emission-reduction measures, which would improve the GTFWUE more effectively. As previously mentioned, urbanization level had a promoting effect on GTFWUE improvement; because of the high water prices in cities, urban residents had a higher awareness about water conservation than rural residents. Therefore, the government should focus on further enhancing rural residents’ water conservation awareness.

    • Water shortages and pollution are major bottlenecks in China’s progress towards sustainable development. The sustainable use of water resources necessitates the improvement of the GTFWUE. This study established an evaluation index system for the GTFWUE, which involved three input indices (labor force, capital stock, and water consumption), a desirable output (economic value), and an undesirable output (wastewater discharge). On this basis, the mSBM was selected to evaluate the GTFWUEs in Chinese provinces from 2000 to 2018. We also investigated the regional differences and spatial correlations of GTFWUEs. Furthermore, spatial panel data models were developed to determine the factors influencing GTFWUE. The main findings of this study are as follows.

      (1) There were significant differences in the provincial GTFWUE values. During the study period, the mean GTFWUEs of only four provinces (Tianjin, Liaoning, Shanghai, and Yunnan) were on the optimal frontier, whereas, those of other provinces did not reach the frontier, indicating that further improvement was required. Most of the provinces with relatively good GTFWUE values belonged to economically developed eastern coastal areas, while most low-GTFWUE provinces were situated in underdeveloped inland areas.

      (2) Similar GTFWUE trends were observed in the eastern, central, and western regions of China during the study period. However, the GTFWUE in the three regions varied significantly; the eastern region had the highest GTFWUE, followed by the central and western regions.

      (3) The Global Moran’s I value indicated that a significant spatial autocorrelation existed among the provincial GTFWUEs. This implied that the GTFWUE values of neighboring provinces were highly similar. According to the LISA maps, most provinces belonged to the H-H and L-L cluster regions, while only a few provinces were in the L-H and H-L cluster regions. Therefore, China’s provincial GTFWUE exhibited a significant clustering effect in local areas and a certain spatial heterogeneity.

      (4) The GTFWUE was promoted by economic growth, population size, opening-up, and urbanization level, and was significantly inhibited by water endowment, technological progress, and government influence. Currently, the water-use structure does not have any significant impact on the GTFWUE, which is closely related to the regional imbalance of agricultural development in China.

      The above findings demonstrated that water use in China remains inefficient, and further efforts are required to achieve sustainable utilization of water resources. Therefore, the following policy recommendations are suggested from the perspectives of cross-regional cooperation, economy, society, and institutions.

      (1) The design of strategies and actions for water utilization and conservation should fully consider the regional differences and prioritize central and western inland areas..

      (2) China should establish a cross-regional system for water resource protection, cooperation, and exchange. The developed provinces in the eastern region must cooperate with underdeveloped provinces in the western and central regions. Moreover, the developed and underdeveloped regions must frequently exchange methods, technologies, management, and experience of industrial, agricultural, and domestic water utilization, to overcome the inadequacies of the western and central regions in water conservation.

      (3) The traditional mode of economic growth should be transformed into a modern high-quality mode. Specifically, China must further promote the opening-up policy and increase access requirements on foreign investment, preventing the investments in projects that might lead to pollution. In addition, China must further improve its urbanization level and consider a new urbanization avenue, comprising intensive planning, construction, and management of urban water resources.

      (4) China should actively enhance public awareness on water conservation. For example, scientific knowledge dissemination and education activities could be organized to keep citizens informed about the severity of water pollution, encourage their participation in water conservation policymaking, and promote their supervision of water management. Moreover, agricultural production must be fully upgraded to ensure maximum water conservation, for example, by promoting advanced techniques such as sprinkler and trickle irrigation. Furthermore, the government must effectively guide the R&D activities of enterprises, promoting them to improve existing technologies, and address the market gaps through R&D. For example, the government can formulate guidelines, such as innovative technical indicators for the R&D of water resource protection products and guidelines for the R&D of water pollution control products.

      (5) China’s water resource management system can be improved by a few methods. First, the government must establish a water-use management system that enhances the enthusiasm and initiative of water management agencies through reasonable water pricing and proper incentive mechanisms. Second, the legal system of water resources and the environment should be upgraded, for example, by establishing laws to enforce strict wastewater discharge standards on enterprises. Third, the government should moderately reduce direct fiscal expenditure and utilize economic instruments (e.g., tax incentives and price subsidies) for market regulation, improving the efficiency of water resource allocation.

参考文献 (82)

目录

    /

    返回文章
    返回