University of Groningen
Quantifying economic-social-environmental trade-offs and synergies of water-supply
constraints
Zhao, Dandan; Liu, Junguo; Sun, Laixiang; Ye, Bin; Hubacek, Klaus; Feng, Kuishuang; Varis,
Olli
Published in:
Water Research
DOI:
10.1016/j.watres.2021.116986
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2021
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Zhao, D., Liu, J., Sun, L., Ye, B., Hubacek, K., Feng, K., & Varis, O. (2021). Quantifying
economic-social-environmental trade-offs and synergies of water-supply constraints: An application to the capital region of
China. Water Research, 195, [116986]. https://doi.org/10.1016/j.watres.2021.116986
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
ContentslistsavailableatScienceDirect
Water
Research
journalhomepage:www.elsevier.com/locate/watres
Quantifying
economic-social-environmental
trade-offs
and
synergies
of
water-supply
constraints:
An
application
to
the
capital
region
of
China
Dandan
Zhao
a,b,
Junguo
Liu
b,∗,
Laixiang
Sun
c,d,e,∗,
Bin
Ye
b,
Klaus
Hubacek
f,
Kuishuang
Feng
c,
Olli
Varis
aa Water & Development Research Group, Department of Built Environment, Aalto University, PO Box 15200, 00076 Espoo, Finland b School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China c Department of Geographical Sciences, University of Maryland, College Park, USA
d School of Finance and Management, SOAS, University of London, London, UK
e Institute of Blue and Green Development, Weihai Institute of Interdisciplinary Research, Shandong University, Weihai, China
f Integrated Research of Energy, Environment and Society (IREES), Energy and Sustainability Research Institute (ESRIG), University of Groningen, Groningen,
the Netherlands
a
r
t
i
c
l
e
i
n
f
o
Article history: Received 3 October 2020 Revised 23 February 2021 Accepted 24 February 2021 Available online 27 February 2021Keywords:
Industrial transition trade-offs synergy
supply-constrained multi-regional input-output (mixed MRIO) model multi-criteria decision analysis (MCDA) sustainable development goals
a
b
s
t
r
a
c
t
Sustainablewatermanagementisoneofthesustainabledevelopmentgoals(SDGs)andischaracterized byahigh level ofinterdependencieswith otherSDGsfromregional to globalscales.Many water as-sessment studiesarerestricted tosilo thinking,mostly focusingonwater-related consequences,while lackingaquantificationoftrade-offs andsynergiesofeconomic, social,and environmentaldimensions. To fillthisknowledgegap,weproposea“nexus” approach thatintegratesawater supplyconstrained multi-regionalinput-output(mixedMRIO)model,scenarioanalysis,andmulti-criteriadecisionanalysis (MCDA) toquantify the trade-offs and synergies atthe sectorallevel for the capital regionofChina, i.e.theBeijing-Tianjin-Hebeiurban agglomeration.Atotalof120industrialtransitionscenarios includ-ingninemajorindustrieswithhighwater-intensitiesandwaterconsumptionundercurrentdevelopment pathwaysweredevelopedtofacilitatethetrade-off andsynergyanalysisbetweeneconomicloss,social goals(here,thenumberofjobs)and environmentalprotection(withgreywaterfootprintrepresenting waterpollution)triggeredbywaterconservationmeasures.Oursimulationresultsshowthatan imposi-tionofatolerablewaterconstraint(anecessarywaterconsumptionreductionforregionalwaterstress leveltomovefromseveretomoderate)intheregionwouldresultinanaverageeconomiclossof68.4 (± 16.0)billion Yuan(1 yuan≈ 0.158 USD$ in2012),or 1.3% of regionalGDP, a lossof 1.94(± 0.18) millionjobs(i.e.3.5%oftheworkforce)andareductionof1.27(± 0.40)billionm3orabout2.2%ofthe
regionalgreywaterfootprint.Atolerablewaterrationinginwater-intensivesectorssuchasAgriculture, Foodandtobaccoprocessing,ElectricityandheatingpowerproductionandChemicalswouldresultinthe lowesteconomicandjoblossesandthelargestenvironmentalbenefits.BasedonMCDA,weselectedthe 10bestscenarioswithregardtotheireconomic,socialandenvironmentalperformancesasreferencesfor guidingfuturewatermanagementandsuggestedindustrialtransitionpolicies.Thisintegratedapproach couldbeapowerfulpolicysupporttoolfor1)assessingtrade-offsandsynergiesamongmultiplecriteria andacrossmultipleregion-sectorsunderresourceconstraints;2)quantifyingtheshort-termsupply-chain effectsofdifferentcontainmentmeasures,and 3)facilitatingmoreinsightful evaluationofSDGsatthe regionallevelsoastodetermineprioritiesforlocalgovernmentsandpractitionerstoachieveSDGs.
© 2021 The Authors. Published by Elsevier Ltd. ThisisanopenaccessarticleundertheCCBY-NC-NDlicense (http://creativecommons.org/licenses/by-nc-nd/4.0/)
∗Corresponding authors.
E-mail addresses: liujg@sustech.edu.cn (J. Liu), lsun123@umd.edu , lsun123@ umd.edu (L. Sun).
1. Introduction
InSeptember2015,193membersoftheUnitedNationsadopted the 2030 sustainable developmentagenda (United Nation, 2016). This agenda features 169 targets under 17 sustainable develop-ment goals(SDGs) in response torapidly rising consumption
de-https://doi.org/10.1016/j.watres.2021.116986
mand, resourcescarcity,environmentalcontaminationandclimate extremes, with the commitments to implement those by 2030 (Nerinietal.,2018).Inaggregate,theseSDGsentailaroadmapto ensure human well-being, economic prosperity and environmen-talprotectionbytacklingmultiplechallengesfacedbyhumankind, of which, Goal 6 (Sustainable water and sanitation) is the one with highshares oftrade-offs andsynergies withother SDGson global and national scales (Pradhan et al., 2017), these interac-tions emphasize the importance of waterresources in determin-ing theachievementofotherSDGs(Szaboetal.,2016).The wors-eningscarcity ofwaterresources hasbecomeathreatto the sus-tainabledevelopmentofmodernsociety,especiallyinsomerapidly developingregionslikeIndiaandChina(GoslingandArnell,2016;
MekonnenandHoekstra,2016; Zhaoetal., 2015;Liuetal.,2017). AvailablewaterpercapitainChinaisonly1,976m3in2018,which
isjustone-fourthoftheglobalaverage(Guetal.,2017;Ministryof WaterResources ofChina, 2018). Therefore,watershortages have become“chokingpoints” thatpotentiallyrestricteconomic produc-tioninChina,especiallyinthewaterscarcenorthernregionssuch as Beijing-Tianjin-Hebei metropolitan region (the so-called “capi-talregion” or“Jing-Jin-Jiregion”).Accordingtothelateststatistics, thecapitalregionisfeeding8% ofChina’spopulation(110million inhabitants) and producing nearly 10% of China’s GDP withonly 0.6% of total wateravailability (18.1 billion m3 in 2017), andthe
freshwater endowment per capita is only 165 m3 (National Bu-reauofStatisticsofChina,2018).Theimbalancebetweeneconomic production anddistribution ofavailable water resources has hin-dered further sustainable development. Toalleviate water stress, in 2012, the central government proposed the “the most strin-gent water resource management system” or so-called “Redline” regulations, which is measured by the “three redlines”: Control-ling national water use, improving water use efficiency and re-ducing wastewater dischargeto conservelimitedwater resources (Liu et al., 2013; Li et al., 2020). As a result, the capital region hasputforwardthe“Jing-Jin-Jiintegrationstrategy” inresponseto thisambitiouswaterstrategyandrisingwaterdemand(TheState Council ofChina,2015).There-evaluationofdevelopmentoptions under the special consideration of water supply constraints was the key feature of this integration policy. Thus, planning future economic developmentfroma demand-sideperspective basedon waterendowmentisasignificantmeasureinthisstrategy.
Available waterresourcesassessments focusmainlyon assess-ing theenvironmental status ofwater resources in terms of wa-ter quantity, quality and scarcity impacts, and rarely consider other environmental aspects (Liu et al., 2017). Tomodel and as-sess the interactions of economic activities and their impacts on water resources at regional, national and global levels, hydro-logical models and water footprint accounting are widely used (Hoekstra et al., 2011; Zhuo et al., 2016; Mao and Liu, 2019;
Qi etal.,2018;Xuetal., 2019;Liuetal., 2020). Alargebunch of globalhydrologicalmodelsforcedbyclimatemodels, greenhouse-gas concentration scenarios and shared socioeconomic pathways havebeendevelopedandintegratedtoassessclimatechange im-pactsonwaterscarcity fromanearthsystemsciencesperspective (Scheweet al., 2014; Wada etal., 2013, 2017; Prudhomme etal., 2014; Wangetal., 2021). Many scholarshave alsointegrated hy-drologicalmodels withhuman activitiesto assessthewater foot-printdistributionanditscontributiontothewaterscarcityat var-ious spatial-temporal scales within the framework ofwater foot-print accounting since the concepts and methodologies of “vir-tualwater” and“waterfootprint” wereintroduced(Hoekstraetal., 2011;Allan,1996).Asaresult,hydrologicalmodelsprovidea com-prehensiveunderstandingonthemechanismsthatshapethe avail-ability, cycling and quality of water in geological and hydrolog-ical terms (Oki and Kanae, 2006), whereas the water footprint approach can be used foraccounting water consumption caused
by economic productionand consumption (Mekonnen and Hoek-stra, 2011; Liuetal., 2015). Inaddition, considerableefforts have beenmadetosimulatetheimpacts ofwateruse.Theapproaches forassessingtheimpacts canbe categorizedintothreemain cat-egories: indicator systems (Schlör et al., 2018), system dynamic models (Zhang et al., 2019; Wang etal., 2019), and input-output (Feng and Hubacek, 2015; Feng et al., 2014) or ecological net-work analysis (Wu et al., 2016; Yang et al., 2012). For instance,
Caietal.(2017)usedacompositeindexapproachtodemonstrate thespatial-temporal characteristicsofChina’swaterresource vul-nerability to highlight key challengesof China’s water resources.
Wang etal. (2019) introduced a comprehensive modeling frame-workbasedonsystemdynamicapproachforintegratedwater re-sources management (IWRM) to provide users with social, eco-nomic and environmental assessments from the perspective of basin-scale watersecurity inthe Bowriver basin ofCanada. Fur-thermore,many scholars have applied input-output or ecological network analysis to calculate virtual water trade across regions andsectorsandquantifythedistributionorallocationofwateruse throughcomplexeconomicactivities(Zhaoetal.,2017;Zhaoetal., 2015; Fang and Chen, 2015; Guan et al., 2014; Hubacek et al., 2009).
Despite previous studies have provided a solid basis for wa-terresourceassessmentsinendowment,vulnerabilityandscarcity, they have frequently ignored trade-offs and synergies between protectingwaterresourcesandotherSDGs.Thishadledtoa grow-ing recognition that water underpins economic andsocial devel-opment, without which it would be impossibleto achieve other SDGs successfully (Bizikova et al., 2013). For instance, achieving water sustainability by reducing demand can lead to trade-offs bothintermsofeconomic output(SDG8,12)andintermsof hu-manwell-being(SDG1, 2,3,4,7,8),butmightgeneratesynergies withenvironmentalprotection(SDG13,14,15).Becauseofthe in-terconnectednatureofwater,economicdevelopment,socialissues and other environmental factors and the aspiration to improve them simultaneously, quantifying the trade-offs and synergies ofeconomic-social-environmental dimensionsunderwater-supply constraints is crucial. The nexus framework has emerged to ad-dresstheinteractionsbetweenenvironmentalresources andmost recentlythelinksamongfood,energyandwatersystemsthrough couplingnetworkanalysis,lifecycleassessmentandfootprint anal-ysis to quantify water and other resource footprints at different scales (Liu et al., 2019; Zhou et al., 2019; Newell et al., 2019;
Kurian,2017; Whiteetal., 2015,2018; Castilloetal., 2019).These methods help to show how these resources or impacts virtually flow through production and consumption networks (Wang and Chen,2016;Feng etal., 2019; White etal., 2018), butare unable to explicitly consider supply constraints and evaluate trade-offs among different indicators.Some scholars have pointed out that the impacts of supply chain disruption may resultin production bottlenecks influencing other sectors and regions via the reduc-tionintheintermediate demands(SahinandOkuyama,2009);in turn,thisbottleneckcouldleadtocascadingeffectsresultingfrom a decline level of activities across the supply-demandchain. For example,HubacekandSun(2001,2005)developedamixed input-output (IO) model featured by land-supply constraints to evalu-atehow thechanging economyandsocietyofChina affectwater useandlanduseattheregionallevelrepresentingprovincesand hydro-economic regions respectively.Liang etal. (2016)proposed an integrated approach based on mixedIO andlinear regression models toestimate the overalleconomic effectandcarbon emis-sionunderelectricityrationingtriggeredby heatwavesin Shang-hai,China.
In general, the quantification of the impacts of supply con-straintsandtheassociatedtrade-offsandsynergiesareofprimary importance in mitigating the vulnerability of modern economies.
Fig. 1. Study area.
However, the existing literature has rarely evaluated such com-pound impacts underresourceconstraints,oftenbeinglimitedby thechosen indicators(e.g.Gallietal., 2011).Morespecifically,the literaturedoesn’ttakeintoconsiderationspillovereffectsofsupply constraints because the studied economy is typically in a closed ratherthanopeningsetting,makingitimpossibletoconsider inter-regional trade flows in the analysis. Actually, the teleconnection of waterconsumption betweenproducersandconsumers accom-panied by interregional trade is increasingly growing along with the expansionoftrade(Caietal.,2020;Zhaoetal., 2015). There-fore,asystematicaccountingofwaterusealongthesupply–chains andtradenetworkwillextendtheopportunitiesfortrades-offsand synergies between economic, social and environmental impacts among differenttradingpartnersfromtheperspective of produc-tionandconsumption(Hubaceketal.,2014).Thismeansthatthere isanurgentdemandforaunifyingframeworkcapableofcapturing trade-offsandsynergiesbetweenwaterandotherSDGsina multi-regionalandmulti-sectoralsetting.Thisresearchaimstoestablish suchaframeworkanddemonstrateitsutility.
In more detail,we propose a ‘nexus’ approachthat integrates a water supply constrained multi-regional input-output (Mixed MRIO)model,scenarioanalysisandmulti-criteriadecisionanalysis (MCDA) toquantify thetrade-offsandsynergies ofeconomic, so-cial and environmental indicatorsunder water-supplyconstraints at the regional-sectoral level. We regard water resources as the bottleneck in the production system. We selected China’scapital regionasourresearchareabecauseoftheseverewatershortages in this region (Fig. 1). First, we developed 120 water-constraint scenarios basedonwaterstresslevel,sectoralwaterintensityand economicimportanceintheregion.Second,wemixedMRIOmodel with water supply constraints to calculate economic losses trig-gered by rationing water usein individual key sectorsin the re-gion,andtoestimatetrade-offsandsynergiesbetweensocial well-being (unemployment)andenvironmental protection(grey water footprint reduction) induced by waterrationing. Third, we
intro-ducedMCDAtoselectoptimalscenarios.Finally,weexplored sus-tainable developmentpathwaysofthe capitalregion andthe po-tentialofthisframeworktoinformandguidepolicy.
2. Materialsandmethods
2.1. Watersupply-constrainedmulti-regionalinput-output(mixed MRIO)model
The classic MRIO model, which takes the form of
(
I− A)
x= f or x=(
I− A)
−1f, is a demand-driven model, that is, thefi-nal demand (f) of the economic sector is an exogenous vari-ableof theeconomic system thatis predeterminedby other fac-tors(consumerpreferences,governmentbehavior,etc.)outsidethe model. Accordingly, a change of final demand (
f) will lead to changes in gross economic output (
x). A MRIO model consists of a systemof linear equations,which describes the distribution of a region-sector’s product throughout the multi-regional econ-omy(MillerandBlair,2009). Thisinterdependenceofregionsand sectorsmakesthe approachpowerfulforassessingthedirectand indirectimpactsofalternativepolicyselectionsacrossregionsand sectors.Additionally,themethodcanbeusedtodealwithshocks andcontingenciesacrosstheeconomy.Policyevaluationbasedon MRIOanalysiscansupplynewinsightstotheeffortsonsearching themostpromisingpolicychoice(Garciaetal.,2020).
To calculate resource consumption (water, energy, land, etc.) triggeredby࢞f,weextendthestandardMRIOmodelwitha diag-onalofsectoralresources requirementcoefficientsmatrixe.Then thesectoral changein resourceconsumption triggered by ࢞fcan becalculatedasfollows:
E=e
(
I− A)
−1f (1) Equation (1) is used to estimate the changes in economic, so-cial andenvironmental impacts induced by waterrationing from consumption perspective.We selectedGDPloss torepresent eco-nomicimpact, employmentloss(unemployment)torepresent so-cialimpact,andgreywaterfootprintreduction torepresent envi-ronmentalbenefits.Thegreywaterfootprintrefersto“thevolume offreshwaterthatisconsumedtoassimilatetheloadofpollutants basedonnaturalbackgroundconcentrationsandexistingambient waterquality standards” (Hoekstraetal., 2011)Thus,thediagonal elementsinvariable eis value-addedper unit of grosseconomic output,thenumberofjobsperunitofgrosseconomicoutputand greywaterfootprintintensityrespectively.
The standard MRIO model assumes that economy adjusts to changes in spending patterns within a given year, and all pro-duction activities are driven by final demand and fully endoge-nous. That is to say, supply is to be elastic in all regions and sectors perfectly, and a change in final demand is sufficient to stimulatechangesinproductionoutputsandincomesacrossother sectors and regions. However, in this case, the situation is ob-viously that water rationing sectors will not expand or shrink its output level automatically in proportion withchanges in
f.
Equation(1)wouldprovidemultipliersthatareunrealisticallylarge duetoanelasticsupplyresponseassumption.
Thus, a supply constrained MRIO model may be appropriate forthisstudyinwhichfinal demandforsomeregion-sectorsand grossoutputsfortheremainingregion-sectorsarespecified exoge-nously.This method isa techniquethat allows theestimation of economic impacts of exogenous changes originatingfrom supply constraints,suchasthosecausedbystrikes,naturaldisasters, pan-demicssuchastheCOVID-19,tradebarriersorresourceshortages (DavisandSalkin,1984;Artoetal.,2015).Toillustratesupply con-straintscausedbywatershortage,wepresentacaseoftworegions (I, J) with two sectors (1, 2). Products produced by each region-sector canbe merchandisedasintermediate inputsorfinal
prod-ucts(seeTableS1),TableS1canbeexpressedasasystemoflinear equations:
XI XJ = ZII ZIJ ZIJ ZJJ 1 1 + YII YJI +YIJ +YJJ (2)WhereZmatrixistheintermediateusebysectorkinregionIand J;YIJisthefinal demandofregionJforgoodsproducedbysector
kofregionI;XI isthetotaloutputofsectorkinregionI,andVI
isthevalue-addedofsectorkinregionI.
The input coefficients are AIJ=ZIJ
(
XˆJ)
−1, and Equation (2) isrewrittenasfollows:
XI XJ = AII AIJ AIJ AJJ XI XJ + YI YJ (3)Then,reorderingEquation(3)yieldsthefollowing:
I− AII −AIJ −A IJ I− AJJ XI XJ = YI YJ (4)In standard MRIO analysis, the gross output is assumed as endogenous variable and the final demand as exogenous,
Equation(4)couldberewrittenasfollows:
⎡
⎢
⎣
xI 1 xI 2 xJ1 xJ2⎤
⎥
⎦
=⎡
⎢
⎢
⎣
1− aII 11−aII 21 −aJI 11 −aJI 21 −aII 12 1− aII 22
−aJI 12 −aJI 22 −aIJ 11 −aIJ 21 1− aJJ 11
−aJJ 21 −aIJ 12 −aIJ 22 −aJJ 12 1− aJJ 22
⎤
⎥
⎥
⎦
−1⎡
⎢
⎣
yI 1 yI 2 yJ1 yJ2⎤
⎥
⎦
(5)We assume that some external shocks (e.g. resources shortage, pandemic etc.)happen in sector 2 ofregion J, which willshrink theproductioncapacityoftheaffectedsector.Theinitialimpactof the external shockis the reduction inthe gross output ofsector 2inregionJ,subsequently,thisinitialreduction willdecreasethe demand forgoods fromtheindustriessupplying intermediate in-puts totheconstrainedsectorboth directlyandindirectly.Inthis case, the gross output of sector 2 in region J should be exoge-nous, whilethefinal demandisendogenous;thus,werearranged
Equation (5) to leave the endogenousvariables on the left-hand side andtheexogenousvariablesontheright-handsidetoobtain thefollowing:
⎡
⎢
⎢
⎣
1− aII 11−aII 21 −aJI 11 −aJI 21 −aII 12 1− aII 22
−aJI 12 −aJI 22 −aIJ 11 −aIJ 21 1− aJJ 11
−aJJ 21 0 0 0 −1
⎤
⎥
⎥
⎦
⎡
⎢
⎣
xI 1 xI 2 xJ1 yJ2⎤
⎥
⎦
=⎡
⎢
⎢
⎣
1 0 0 0 0 1 0 0 0 0 1 0 aIJ12 aIJ22 aJJ12 −1− aJJ 22⎤
⎥
⎥
⎦
⎡
⎢
⎣
yI 1 yI 2 yJ1 xJ2⎤
⎥
⎦
(6)byarrangingEquation(6),wefindthefollowing:
⎡
⎢
⎣
xI 1 xI 2 xJ 1 yJ2⎤
⎥
⎦
=⎡
⎢
⎢
⎣
1− aII 11−aII 21 −aJI 11 −aJI 21 −aII 12 1− aII 22
−aJI 12 −aJI 22 −aIJ 11 −aIJ 21 1− aJJ 11
−aJJ 21 0 0 0 −1
⎤
⎥
⎥
⎦
−1⎡
⎢
⎢
⎣
1 0 0 0 0 1 0 0 0 0 1 0 aIJ12 aIJ22 aJJ12 −1− aJJ 22⎤
⎥
⎥
⎦
⎡
⎢
⎣
yI 1 yI 2 yJ 1 xJ2⎤
⎥
⎦
(7)Finally, thismixedMRIO modelwithsupply constraintscould be expressedasfollows:
Xno Fco = P(k×k) 0(k×(n−k)) R((n−k)×k) −I((n−k)×(n−k)) −1 × I(k×k) Q(k×(n−k)) 0((n−k)×k) S((n−k)×(n−k)) × ¯Fno ¯Xco (8)Thesub-matricesinEquation(8)aredefinedasfollows:
P(k×k) is the k × k matrix that is extracted from the first k columns andkrowsofmatrix(I-A), andrepresentstheaverage
expenditurepropensitiesoftheunconstrained sectorsinthe sup-plyside. The firstksectorsare the endogenousandthelast (n -k) sectorsare the exogenous ones. R((n−k)×k) is the(n -k) × k matrixfromthefirstkcolumnsandthelast (n- k)rowsof(-A), which is the average expenditurepropensities of the non-supply constrainedsectorsonthesupplyconstrainedsectors.
Q(k×(n−k))isthematrixfromthefirstkcolumnsandlast(n -k)rowsofAmatrix,andrepresentstheexpenditurepropensitiesof supplyconstrainedsectorsonthenon-supplyconstrainedsectors.
S((n−k)×(n−k))ismatrixthat isextractedfromthelast(n-k) rowsandcolumnsof-(I-A),anditrepresentstheaverage expen-diturepropensitiesamongthesupplyconstrainedsectors.
¯
Fno is columnvector of elementsfromy1 toyk,which means
the exogenousfinal demandfor theunconstrained sectorsin the supplyside.
¯
Xcoiscolumnvectorofelementsxk+1throughxn,whichmeans
the exogenous total economic output for the supply constrained sectors.
Xnois column vectorwithelementsx1throughxk,representing
theendogenoustotaleconomicoutputofunconstrainedsectorsin thesupplyside.
Fco is column vector with elements yk+1 through yn,
repre-senting the endogenous final demand of the supply constrained sectors. n is the numberof sectors in the IO table, and k refers to the number of water rationing sectors. Equation (8) can be easily converted in a difference form, with
X¯co=X¯co0 − ¯Xco, i.e.,
theconstraintinduced output reductionin comparisontothat in the reference economy (X¯0
co) without the imposition of the
con-straint, which will be determined by the scenario calibration in
Section 2.2, and
F¯nore f ersto the change of the final demand
caused by exogenous water shortage, mixed MRIO model is em-bedded inan opening market, allowing forimports fromoutside theregiontocompensateitsshrinking finaldemand.Thuswe as-sumethat there is noexogenous change inthe final demandfor non-constrained sectors (
F¯no=F¯no0 − ¯Fno=0
)
, meaning that F¯noremainingthesameasinthereferenceeconomy(F¯0
co)withoutthe
impositionofwaterrationing.
Xnocorrespondsto thechangein
the economic output of the non-water supply constraint sectors triggeredbytheindirectimpactofthewatersupplyconstraint sec-tors.
Fcorepresentsthechangeinthefinal demandofthewater
supplyconstraintsectors. Finally,wecan calculatetheimpacts in economic, social andenvironmental dimensions from production perspectivebasedonEquation(8)throughpre-multiplyeby
Xno
and
X¯co. Hubacek and Sun (2005) described how to formulate
thismodel(HubacekandSun,2005).
2.2. Industrialtransitionscenariosforevaluatingwaterrationing acrosssectors
Thesustainabilityconsiderationindetermining production un-derwatersupplyconstraintisthat economicdevelopmentshould be aligned with the carrying capacity of local water resources. Tradeallowslocalshortagestriggeredbywaterconservationtobe offset.Toensurelivelihoodsandtomeetecologicalwater require-ments,theallocationofwaterresourcesacrossdifferenteconomic usesisanimportantmeasuretoinformdemand-sidemanagement. Thus, industrial transition scenarios are designed to reduce wa-terconsumptioninproductionactivities.Inthisstudy,wedefined “watersupplyconstraints” aswaterconsumptionthatmustbe re-ducedto mitigateregional waterstress by one level inlinewith waterstressindex,whichiscategorizedbasedontheratioofwater consumption towateravailability forhuman,thisindexoriginate from Hoekstra et al.(2012) and Mekonnen and Hoekstra (2016), manyprevious publicationshasapplied thisindexfortheir stud-ies(e.g.Chouchane etal., 2020;.Maetal., 2020;Zhuoetal.,2016;
envi-Table 1
Designing water supply constraints in the capital region in 2012.
Row NO. [Billion m 3 ] Beijing Tianjin Hebei Data sources
1 Total available water availability 3.95 3.294 23.553 ( Ministry of Water Resourcesof China., 2012 ) 2 Water availability for human (0.2 ∗Row1) 0.79 0.659 4.711
3 Total water consumption 1.96 1.55 14.4 ( Ministry of Water Resources of China., 2012 )
4 Total water consumption in economic sectors 0.95 0.984 15.742 ( Ministry of Water Resourcesof China., 2012 ; Zhao, 2019 ) 5 Current water stress level (%) (Row 3/ Row 2) 248.1 235.3 305.7
6 Water stress category Severe Severe Severe
7 Targeted level (%) 200 200 200
8 Targeted category Significant Significant Significant 9 Required reduction in water consumption
((Row 7 – Row 5)/100 ∗Row 2) -0.38 -0.232 -4.979 10 Required reduction water consumption in
economic sectors (Row 4/ Row 3 ∗Row 9) -0.184 -0.148 -5.443 11 The contribution of adjustments in industrial
structure (%)
21 21 21
12 Water supply constraint (Water gap) (Row 10
∗Row 11/100) -0.039 -0.031 -1.143
ronmentalwaterrequirement–regardedas80%oftotalwater avail-ability is thesimplest way(Hoekstra etal., 2012,Mekonnen and Hoekstra (2016)). Severalstudies haveshownthatchanges tothe industrial structure havebecome the biggest decelerator in driv-ingwaterconsumptioninsome regions(Liuetal.,2018;Mietal., 2017;Planketal.,2018;Zhao,2019;Caietal.,2016).Forexample,
Zhao (2019) using structuraldecomposition analysisshowed that structuralchange intheeconomy ofBeijingledtoa reduction in waterconsumption by21%between2002and2012.Thus,we as-sumethat21%ofreducedwaterconsumptioninTianjinandHebei could alsobe achieved through adjusting theindustrial structure in the future (Zhao, 2019). Table 1 shows the step by step pro-cessofhow we couldobtain (tolerable)watersupply constraints. Finally,the(tolerable)watersupplyconstrainttranslatesintoa re-duction ofthewaterconsumption by0.039 billion m3 (2%ofthe
total)fromthereference-yeareconomyforBeijing,0.031billionm3
(2%ofthetotal) forTianjinand1.143(8%ofthetotal) billionm3
forHebei(seeTable1).
After determiningthe necessaryreduction toachieve tolerable water constraints based on water stress levels in the capital re-gioninthebaseyear,thenextstepwastodeterminethesectoral distribution of the reduction. Referring to the “Measures for in-dustrialstructureadjustmentwithenergyconservation” (Songand Liu, 2013) and available literature related to the water footprint at the sectoral scale we designed industrial transition scenarios (Zhao et al., 2017). In order to mitigate the adverse impacts of watershortageson humanhealthandwell-being,tertiary sectors suchasFreighttransportandwarehousing;Hotels,foodand bever-ageplaces(establishments)wouldnotbeconsideredaswater sup-plyconstraintsectors(Liangetal.,2016).Tomaintainthesafeand necessary functioning ofurban systems,the water supplyto key sectorssuch asGas andwaterproductionandsupply;Social ser-vicesandWholesaleandretailwouldnotberationed(Liangetal., 2016).Asaconsequence,thewaterconstraintsaremainlyimposed uponagricultureandsomesecondarysectors.
Our previous research results showed that some sectors such as Agriculture, Food and tobacco processing, and Textile indus-trieshavehigherwaterdependencyintermsofdirectwater con-sumptionandwaterfootprint(Zhao,2019;Zhaoetal.,2017);thus, we identified the top five non-service sectors with highest di-rect water consumption and water footprint as the water sup-ply constraintsectors tocalculate thetrade-offs andsynergies of economic-social-environmental aspects triggered by water short-age.Agriculturalwaterconsumptionaccountedformorethan60% of total waterconsumption dueto the specialproduction law of agriculture in form of evapotranspiration (ET) during the whole growingperiod(Zhao,2019),whichhadaverylargeimpacton
wa-Fig. 2. Flowchart of the evaluation procedure.
ter sustainability. We regarded Agriculture asa mandatory water supplyconstrainedsector,andotherselectedsectorsasadditional constrainedsectors. Weallocatedthewatersupplyconstraint(see
Table1)acrosstheselectedsectorsaccordingtotheir proportions ofcurrentwaterconsumptioninthetotal,andthenwecalculated the direct decline ofeconomic output caused by the hard water availability constraints,based on waterintensityper unit of eco-nomicoutput.Finally,wedeveloped3scenarios forBeijing,5 sce-nariosforTianjinand8scenariosforHebei(seeTable2),andused asupplyconstrainedMRIOmodel tocalculatethetotal economic loss across all sectors. At the regional scale, the combination of scenariosusedtomeettheimposedwatersupplyconstraintinthe capital region could be “B1+T1+H1” or “B3+T4+H7” in Table 2. Consequently,thetotalcombinationsthatmeettheconditionwere equal toC13× C1
5× C18=120. Fig. 2 presents the flowchart of this
evaluationprocedure.
2.3. Optimalscenarios/pathwaysbasedonmulti-criteriadecision analysis(MCDA)
MCDA helpsbalance multiplecriteria ina structured way. Al-lowingdifferent preferences forthe criteriais particularly signif-icant for policy decisions, where many assessment criteria and, even more frequently, opposing views of different stakeholders, coexist. Scholars have developed numerous MCDA approaches
Table 2
Industrial transition scenarios based on the (tolerable) water supply constraint.
Beijing Tianjin Hebei
ID Adjusted industrial sectors Decline in economic output (10 9 yuan) ID Adjusted industrial sectors Decline in economic output (10 9 yuan) ID Adjusted industrial sectors Decline in economic output (10 9 yuan)
B1 1.Agriculture 2.23 T1 1.Agriculture 1.39 H1 1.Agriculture 37.4
B2 1.Agriculture + 6.Food and tobacco processing
(2.19, 5.69) T2 1.Agriculture + 6.Food and tobacco processing
(1.37, 7.56) H2 1.Agriculture + 4.Metals mining and dressing (Cleaning)
(36.7, 32.2) B3 1.Agriculture + 22.
Electricity and heating power production
(2.0, 17.53) T3 1.Agriculture + 12.
Chemicals (1.33, 8.08) H3 1.Agriculture + 6.Food
and tobacco processing (37.3, 26.2)
T4 1.Agriculture + 14. Smelting and pressing (processing) of metals (1.32, 17.1) H4 1.Agriculture + 7.Textile industry (37.4, 12.1) T5 1.Agriculture + 24. Construction (1.37, 11.17) H5 1.Agriculture + 8. Garments, leather, furs, down and related products
(37.3, 10.4)
H6 1.Agriculture + 14. Smelting and pressing (processing) of metals
(36.9, 88.6) H7 1.Agriculture + 22.
Electricity and heating power production
(36.0, 17.5) H8 1.Agriculture + 24.
Construction
(37.1, 38.7)
(for reviews see e.g. Kumar et al., 2017; Wang et al., 2009). In this research, we adopted the most widely used multi-attribute value theory (MAVT) and multi-attribute utility theory (MAUT) (Wang et al., 2009). We set up three kinds of weights to as-sess the importance of each trade-off and synergy in the con-text of the capital region: priority to the social dimension (0.3 for the economy, 0.5 for society and 0.2 for the environment), priority to economic and social dimensions (0.4 for the econ-omy, 0.4 for society and 0.2 for the environment) and equal weights (0.333 for each). Then, weightedaverages were used to evaluate the scenario rankings (Afganand Carvalho, 2008; Begi´c and Afgan, 2007). Santoyo-Castelazo and Azapagic (2014) and
IshizakaandNemery(2013)provideddetailedprinciplesonthese twomethods.Thekeyequationsareasfollows:
V
(
s)
= 3 i=1 wiv(
s)
i (9) U(
s)
= 3 i=1 wiu(
s)
i (10) Where:V
(
s)
isthecompositevaluefunctionbasedonMAVT, represent-ingthetotalperformancescoreforscenarios;wiistheweightforcriterion i, and threecombinations of wi (0.3/0.5/0.2,0.4/0.4/0.2, 0.333/0.333/0.333) wereusedseparatelyinthisstudy;v
(
s)
iistheranking value that reflects the performance of scenario s on cri-terion i using a scale from1 to 120, where1 is the best option and120istheworstoption.Correspondingly,U
(
s)
isthe compos-ite valuefunction basedonMAUT,u(
s)
iisthesimulatedvalue ofscenario s on criterion i, which has been normalized between0 and 1, where 1 is the best and0 is the worst. We selected the weightedaveragevalue asthethresholdtoevaluatescenario sus-tainability(Liangetal.,2016).
2.4. Datasources
To calculate the trade-offs and synergies of economic-social-environmental pillars at sectoral level based on Equations (1)
-(8), we need the MRIO table, water consumption, labor input,
grey water footprint and value-added or GDP in capital region and the rest of China (ROC) (see Table S3 and Table S4). We collected the provincial level MRIO table for the year 2012 in China from Mi et al. (2017), and therefore we use this table as the reference economy, the information was converted to 2010 constant prices. Forthe focus of thisresearch, the provinces be-yond the capital region were aggregatedinto “Rest of China” re-gion. Water consumption data came from the provincial water resource bulletin (Ministry of Water Resources of China, 2012),
Zhao(2019)andZhaoetal.(2019).Wedistributedirectwater con-sumptionin2012toMRIOsectorsbasedonpreviouswatersurvey data, economic output and some technical assumptions, the cor-responding validation assessment and the workson how to dis-tributewaterconsumption datatomatchwithMRIOsectorshave been done by Zhao et al.(2017) and Liu (2016). Labour datafor each sector at the regional scale were obtained from the China LabourStatistical Yearbook(NationalBureauofStatisticsofChina 2013a),theChina StatisticalBook (NationalBureauofStatisticsof China2013)andtheChinaRuralStatisticalYearbook(National Bu-reau of Statistics of China 2013b). We calculated the grey water footprint based on Hoekstra et al. (2011), Hoekstra and Mekon-nen(2012)andZhaoetal.(2016),whichiscalculated asthe vol-ume of water that is required to dilute pollutants to an extent thatwaterqualityremains aboveagreedwaterenvironment stan-dards.Thechemicaloxygendemand(COD)andammoniacal nitro-gen(NH3-NorNH+4-N)dischargeinwastewaterwere selectedas
waterpollution indicatorstoestimatethegreywaterfootprintfor each sector, the relateddata isretrieved fromthe environmental statistical yearbook(NationalBureau of Statistics of China, 2013) and Pollution Census Dataset (Editoral Board of First Pollu-tion Census, 2011). Sectoral value-added (GDP) is obtained from
Mietal.(2017).
2.5. Limitationsanduncertainties
Some limitations and uncertainties should be included when interpretingtheresults.First,oursupply-constrainedMRIOmodel accountedfortradeflows acrossthe threetarget regions andthe
rest of China, but does not pay much attention to the evalua-tionofthespillovereffectstointernationalmarketsorother coun-triesbecausesuchanextension goesbeyondthescopeofthis re-search. Tocarryout sucha comprehensiveextensionalong global supply-chains,atwo-tierMRIOmodelthatcouplesourMRIOwith a global MRIOwould be needed. Giventhat the focusof our re-search isontheimpactsinducedbywaterrationingatthecapital regions and potential spill over to other regions in China rather than globalregions, we deem thenational level MRIOto be suf-ficient forthe purposes ofthis paper.Second, only one indicator wasselectedforeacheconomic,socialandenvironmental dimen-sion to showhow the framework works.Inthe future,it willbe straightforwardtoaddmoreindicatorsreflectingdifferentSDGsto extendouranalysisusingthisframework. Third,inthisstudy,we attributed regional-invariant weights to economic development, social well-being and environmental enhancement. It would be more policy-relevant to assign weights for each criterion based on regional endowments and stakeholder inputs (Kumar et al., 2017; Cinelli et al., 2014). Fourth, to some extent, social wellbe-ing (herenumberofjobs)istightly linkedto GDP,whereas other SDGs or social targetsare lesswell capturedby an input-output framework.
Fifth,theChinesegovernmentpublishesinput-outputtables ev-ery 5 years, our analysis is based on the economic structure in 2012, butthe technicalmatrixandthus the productionfunctions will gradually changeovertime. Tradingpatternsfrom2012 can-not capturefuturetrends,butforthepurposesofthispaperit is sufficienttokeepthemconstant toshowtheimpactssupply con-straints would have everything else kept constant. It should be also noted that Zheng et al. (2020) constructed the 2015 China MRIO table forinvestigating theregional determinantsof China’s CO2 emission.However, theinterpretation of this2015 MRIO ta-blearedoneintheabsenceofprovincial2015IOtablesandbased on a minimization of an entropy function subject to constraints ofmanyassumedrelationshipson productionstructureand trad-ing patterns.Given thefocus of our research on the response of theintra-provincial industrialstructuretowatershortageand de-veloping oneintegratedmethodframeworktoquantify economic-social-environmental trade-offsand synergiesunder water-supply constraints andits potential applications to support environmen-tal policies.Such a purely calibrated2015 MRIO withoutthe real intra-provincial input-output interactions across industries is not suitableforourresearch.Therefore,weoptedtousethebest avail-able MRIO of Mi et al. (2017) and set the base yearin 2012. It is worth noting that the exact data on which the applicationis based isnotofmajor importance,oncebetter databecome avail-able,thesecanbeeasily usedtoupdatethemodelandapplythe framework. In addition, the selection of pollutants will have an influence on grey waterfootprint to some extent. We selecttwo common pollutants(COD andNH3-N) inthe discharged
wastew-ater as water pollution proxies to estimate the grey water foot-prints for each sector. On the one hand, this selection is largely determinedbythefactthatthesetwopollutantsarethemost fre-quently monitored andrecorded in China andhave been widely employed toevaluate surface waterquality (e.g. Ma etal., 2020;
Zhao et al., 2016; Guan et al., 2014). On the other hand, these two pollutants accounted for more than 84% of total discharged pollutants in agriculture, industry and domestic sectors in 2012 (National Bureau of Statistics of China, 2013). Finally, the reduc-tion of final demand in theregion would inevitablylead to pro-ductionincrease inotherregions tomeetthedemand,the reduc-tion within theregion doesnotnecessarily leadto an overall re-duction in China because of outsourcing, unless we know where the resources mightbeoutsourcedto andthat region’seconomic structure,technologyandwateruseefficiencyaswellasassociated pollution.
3. Results
3.1. Economicloss,employmentlossandenvironmentalgainsby scenarioandregion
Fig.3,Fig.S1andTableS5illustrateeconomicloss(GDP loss), employmentloss(unemployment)andenvironmentalbenefit(grey waterfootprintreduction) triggeredby hypotheticalwatersupply constraintsimposedonChina’scapitalregion.Theintroductionof water rationing in Agriculture (Scenarios B1, T1 and H1) would lead to the least reduction of GDP, which was 1.67 billion Yuan (0.90billionwithinthecapitalregionand0.77billionintheROC), equalto0.1%ofBeijing’sGDP,1.15billionYuan(0.67billionwithin the capital region and 0.48 billion in the ROC), 0.1% of Tianjin’s GDP, and35.5 billion Yuan(22.7billion within thecapital region and12.8 billion inthe ROC), 1.5%of Hebei’s GDPrespectively. In comparison,thehighestlossofGDPwasfromscenarios B3 (Agri-culture + Electricityandheatingpower production), T4 (Agricul-ture+Smeltingandpressing(processing)ofmetals)andH6 (Agri-culture + Smelting and pressing (processing) of metals), which were 10.8billion Yuan(3.75 billion within capitalregion and7.0 billionintheROC),equalto0.7%ofBeijing’sGDP,10.6billionYuan (4.07billionwithinthecapitalregionand6.56billionintheROC), equalto0.9%ofTianjin’s GDP,and89.4billion Yuan(41.7 billion withinthecapitalregionand47.7billionintheROC),equalto3.7% ofHebei’s GDP, respectively. Thedifferencesin GDPlossesby re-gionbetweenscenarios changedfrom6.5timesinBeijing(B3/B1) to2.5times(H6/H1)inHebei,whichindicatesthattheoutput re-sponseofnon-agriculturalindustrialproductstooneunitofwater reduction waslarger thanthat of agriculture,andeach manufac-turesectorhaddifferentextentsofoutputresponsebecauseof dif-ferencesinproduction recipesandprocesses. Ourresultsshowed thatwaterconstraintsreducedeconomicactivitiesnotonlyinthe rationedregions andsectorsbutalsoinotherregions andsectors across the supply-chain upstream and downstream. Forexample, inScenarioH6,53% ofeconomiclosswasfromsupply-chain sec-tors (Freight transport and warehousing, Other services etc.) re-latedtosector1(Agriculture)andsector14(Smeltingandpressing (processing)ofmetals),thus,anexogenousshocksuchaslimitsto wateravailabilitywouldaffectthewholeeconomythrough supply-chainlinkages.
Similartoeconomicloss,onlyadjustingagriculturalwater con-sumption(Scenario B1,T1andH1)wouldhavethelowestimpact on unemployment, namely 0.11 million jobs(0.10 million within the capital region and 0.01 million in the ROC), accounting for 0.76%oftotalemploymentinBeijing,0.04millionjobs(0.036 mil-lion within the capitalregion and0.006 million in theROC), ac-counting for 0.6% of total employment in Tianjin and 1.43 mil-lion jobs (1.265million within capital region and0.17 million in the ROC), accounting for 4.2% of total employment in Hebei. In contrast,theScenarioB3(1.Agriculture+22.Electricityand heat-ingpowerproduction,0.20million),T4(1.Agriculture+ 14.Smelt-ing and pressing (processing) of metals, 0.14 million) and H6 (1.Agriculture + 14.Smelting andpressing (processing)of metals, 2.04 million) wouldlead to the largest losses in employment in this region. The differences in job losses triggered by water ra-tioningacrossscenariosbyregionchangedfrom1.4timesinHebei (H6/H1)to 3.3 times (T4/T1in Tianjin, which were smaller than thatofGDPbecausefarmersaccountedforratherlargeproportion ofthetotalemployment.
Intermsofenvironmentalbenefits,we foundthat ScenarioB3 (1.Agriculture + 22.Electricity and heatingpower production), T4 (1.Agriculture + 14.Smelting andpressing (processing) of metals) andH6 (1.Agriculture + 14.Smeltingand pressing (processing)of metals)contributedtothelargestreductioninthegreywater foot-print,withvalues of0.25billion m3 (0.017 billion m3 within the
Fig. 3. Economic loss, social loss and environmental benefit under water supply constraint. Legend represents sector IDs in Table S3.
capital region and0.235 billionm3 in theROC), equalto 2.0% of
thetotalgreywaterfootprintofBeijing,0.27billionm3(0.046
bil-lionm3 withinthecapitalregionand0.22billionm3 intheROC),
equalto1.9%ofthetotalgreywaterfootprintofTianjin,and1.83 billion m3 (0.25billionm3 withinthecapitalregion and1.57
bil-lionm3intheROC),equalto5.8%ofthetotalgreywaterfootprint
ofHebei.Restricting onlyAgriculture(B1,T1andH1) wouldhave the leasteffect onenvironmental protection,whichwas0.03 bil-lion m3 inBeijing, 0.02billion m3 in Tianjinand 0.45billionm3
inHebeirespectively. Thedifferencesinenvironmentalgains trig-geredbywaterrationingacrossscenariosby regionvariedfrom4 timesinHebei(H6/H1)to16times(T4/T1)inTianjin,whichwere much biggerthaninthecaseofGDPandsociallossesbecauseof significant water pollution dischargein manufacturing compared with agriculture.Our resultsindicate that water-rationing in sec-tors withhigh grey water footprints per unit of output, such as Electricity and heating power production, Smelting and pressing (processing)ofmetals,andChemicals,wouldhavethelargest con-tributiontogreywaterfootprintmitigation.
3.2. Economicandsociallossesversusenvironmentalgainsbysector andregion
The heat map inFig.4andTable S6show the averagerelative sharesandabsolutevaluesinlossofvalueaddedandjobs,and
en-vironmentalbenefitsunderwatersupplyconstraintsatthesectoral scaleinthecapitalregion.Theeconomiclossesvaryfrom36.4%(or 24.9billionYuan)ofthetotalvalueaddedinAgriculture(sector1) to0.4%(0.27billionYuan)oftotalvalueaddedinNonmetal miner-alsmininganddressing(sector5)inthecapitalregion.Thesecond largest onesare fromOther services(sector 30, 12.6% of the to-talvalue added,with8.6billionYuan)andSmeltingandpressing of metals(sector 14, with5.2% of thetotal value added,3.5 bil-lionYuan).Attheregionalscale,Agriculturecontributedwith17% the greatestshare to total economic loss inBeijing, 13% in Tian-jinand41%inHebeirespectively.Similarly,theproportionsofjob lossesrangedfrom77.3%(1.50millionjobs)inAgricultureto0.07% (0.0014million jobs)inOther manufacturingproducts (sector21) of all jobs due to the water constraint, the second largest ones are from Wholesale and retail trade (sector 26, 3.8%) andOther services (3%). At the regional scale, Agriculture still showed the biggestcontribution,rangingfrom70.5%ofalllostjobsinBeijing, 48.4%inTianjinand79.6%inHebeirespectively.
Interms ofenvironmental benefits,Hotels, foodandbeverage places(sector27)hasthelargestgreywaterfootprint(0.541billion m3),accountingfor42.5%ofthetotalgreywaterfootprintin
capi-talregion,thisratiois43%inBeijing,36.6%inTianjinand43.5%in Hebei.Other services (sector30)andChemicals (sector12)show the second largest relative effects, which are 12.8% and 8.3% re-spectively, butsector 20 (Measuring instrument& machineryfor
culturalactivity&officework manufacturing)isthesmallestone, only0.026%ofthetotalgreywaterfootprint.
Yet, when looking at distribution of costs and benefits, some interestingpatternsemerge.Damagestotheeconomyandjobsare heavilyconcentratedinfood-relatedsectors(Agriculture,Foodand tobacco processing(sector6))andservicesectors(Wholesale and retail trade,Other services), however, thebenefit to the environ-ment is mainly dueto sectorswithheavy wastewater discharges (Hotels, foodandbeverageplaces,Chemicals).Forexample,under the watersupplyconstraintsinourdesignedscenarios,the resul-tant economic loss,job loss,andenvironmentalgain insector 30 (Otherservices) inthecapitalregions asawhole willaccountfor 13%, 3%, and13%of thetotal economicand jobloss, andthe to-tal environmental gain, respectively. By contrast,the correspond-ing sharesinsector 27(Hotels, foodandbeverage places)willbe 3%,2%,and43%,respectively.Thismismatchshowsthehotspotsof pollutionaswellaspointsforinterventionatleastcosts consider-ingimpactsalongtheentiresupplychain.
3.3. Industrialtransitionscenariosunderwaterconstraints
Fig.5showstheboxplotsofthe120industrialtransition scenar-iosunderwatersupplyconstraintsinthecapitalregion.The over-all economicloss variesfrom38.3billionYuan(B1T1H1) to110.8 billion Yuan (B3T4H6), andapproximately70% of thesescenarios haveeconomiclossesrangingbetween50billionYuanand80 bil-lion Yuan. Industrial transitions resulted inan average economic lossof68.4(± 16.0)billionRMB,accountingfor1.3%oftotalGDP ofthecapitalregion(NationalBureauofStatisticsofChina,2013). In terms of unemployment, job losses ranged from 1.59 million (B1T1H1)to2.38million(B3T4H6),withanaverageof1.94(±0.18) million jobslost.Thisvalueaccountsfor3.5% oftheregion’stotal employmentin2012(NationalBureauofStatisticsofChina,2013). Meanwhile, the reduction intotal greywater footprintfluctuated between0.49 billionm3 (B1T1H1) and2.34billion m3 (B3T4H6),
which wasnearly 5 timesthe difference betweenmaximum and minimumvalues,andtheaveragereductioninthegreywater foot-print was1.27(±0.40) billionm3,approximately2.2%ofthetotal
greywaterfootprintinthisregion.
3.4. Trade-offsandsynergiesofeconomic-social-environmental dimensions
Fig. 6 shows the 3-dimensional (3D) scatter plot of the 120 scenario combinations. This 3D scatter plot is designed to pro-vide instructions on the future development pathway selection. Wefoundthatwatersecurityhastrade-off relationshipswith eco-nomicgrowthandjobs,andsomemeasuresinducedbywater con-servationwouldbringabouteconomiclossesandincreasesin un-employment; conversely, synergic connections exist between wa-ter conservation policies and the grey water footprint. For the capital region,scenarioswithsectorsofAgriculture, Foodand to-bacco processing and Chemicals (B1T1H1, B2T1H1, B2T3H1, yel-low spheres in Fig.6 etc.)wouldhavelessimpacts on the econ-omy than scenarios with the sectors of Electricity and heating power production, Smelting and pressing (processing) of met-alsandConstruction(B3T4H6,B3T5H6,goldenspheres,etc.). Simi-larly,someadjustedscenariosincludingFoodandtobacco process-ing (B1T1H1, B2T2H1, yellow spheres, etc.) would lead to fewer job losses than scenarios with the sectors Electricity and heat-ing powerproduction,Smeltingandpressing(processing)of met-als (B3T4H6, B3T5H6,golden spheres, etc.). As for environmental benefit,scenarios includingSmelting andpressing (processing)of metals, Electricity and heating power production and Construc-tion (B3T4H6, B3T5H6, B3T3H6, golden spheres, etc.) would mit-igatethegreywaterfootprintmuchmorethanscenarioswiththe
sectors of Food andtobacco processing, and Chemicals (B1T1H1, B2T1H1,B1T2H1,B1T3H1,yellowspheres,etc.).
3.5. Optimalscenariosbasedonmulti-criteriondecisionaid(MCDA)
WeappliedtheMCDAframeworktocalculateacompositevalue astheevaluationindexto measurethe performance ofeach sce-nariointhecontextofeconomic,social,andenvironmental dimen-sions(seeFig.7).InMAVTsystem,theweightedaveragevaluewas 60.5,andscenarioswhosevaluesweresmallerthan60.5were re-gardedassustainableinthethreedimensions;incontrast, scenar-ioswithhighervalueswereconsidered unsustainable.Ourresults showthatadjustingtheAgricultureandFoodandtobacco process-ing sectorsin Beijing, theAgriculture, Foodandtobacco process-ing orChemicals sectorsinTianjin,andtheAgriculture and Elec-tricityandheatingpowerproductioninHebeiwouldhaveamore positiveinfluenceonthedevelopmentofthecapitalregion. Corre-spondingly,inMAUTsystem,theweightedaveragevaluewas0.54, which means that scenarios with values greater than 0.54 were acceptedasreferencescenarios forfuturedevelopment, and sim-ilar toMAVT,changing theshares ofproductionin thesectorsof Agriculture, Food andtobacco processing, Electricity andheating powerproductionandChemicalswouldbelikelytohavethe low-estlossesandthemostenvironmentalbenefits.Basedon compos-itevalues undertwo MCDAtheories,thesetop 10 scenarioswith regardtotheirperformancesintermsofeconomic,socialand en-vironmentalsustainabilitycouldserveasreferencesforguiding fu-tureindustrialtransitionpolicies.
4. Discussion
Weproposed a policysupport techniquethat combinesmixed MRIOmodelswithMCDA theory toevaluatethe consequencesof waterconstraints, andapplied itto the largesturban agglomera-tion in China, the capital region or theJing-Jin-Ji region. On the onehand,theseresultson therankandmagnitudeoftrades-offs and synergies provide useful information for policy makers and planners toeffectively identify prioritypolicy selections and bal-ance policy actions in line with policy considerations and value judgements (Kurian, 2017). On the other hand, our mixed MRIO approach coupled with the MCDA allows to evaluate not only thedirect impacts ofalternativepolicy options butalsothe indi-rect impacts induced by supply-chain effects triggered by exoge-nousshocks (e.g. resource shortage, naturaldisasters, emergency eventsetc.), whichhavemoreadvantagesthanfootprint account-ingorsystemindicatorsinimpactassessments(Liangetal.,2016;
EiserandRoberts, 2002;Leung andPooley,2001). Incomparison withthe results simulated by a standard MRIO modeldriven by consumption activities (i.e. Fig.4, Figure S1 versus Figure S2 and TableS5 versus TableS7), themixedMRIO modeldriven by pro-duction activities under the constraintof resourcesupply is able toprovideamorerealisticrepresentationofstructuralinteractions acrossregionsandsectorsinresponsetoresourceconstraints,and thusbetterfacilitatingthedesignofeconomicdevelopment path-ways. Despite the persistent concern in the literature on water scarcity,there hasbeen a lack ofdevelopmentof simpleand ef-fectivetoolsthatcanbeusedtoclearlyassessthelossesand ben-efitscausedby waterconservationmeasures acrosseconomic, so-cial, and environmental systems. Our research fills an important nicheinwaterresourcefield.
Agriculture is a crucial but high water demanding sector. It mainlyprovides rawproducts(wheat,rice, soybean,cotton),with loweconomicvalue-added,butmanifestshigh-waterrequirements comparedwithmanufacturingindustries.Forexample,inscenarios B1,B2andB3(seeFig.3row1),waterrationingreducesthesame amount ofwater consumption, butGDP lossesin B2 and B3 are
Fig. 4. Proportions in economic and social losses (trade-offs) versus environmental gains (synergy) by sector and region.
Fig. 5. Boxplots of industrial transition scenarios under water constraints.
biggerthaninB1,becausescenarios B2andB3 includeindustries withhigherGDPproductivityperunitofwater,whichmeansthat the elasticityofeconomicoutput towateravailability islarger in non-agriculturalsectorsthanforagriculture.Inotherwords, reduc-tioninvalueaddedfromagricultureislowerthaninothersectors underthesamewaterreductionscenarios.Inaddition,agricultural production islabor intensive, andapproximatelyone-third ofthe labor force is still engaged inagricultural workwith low income (farmers,migrantworkers)inthecapitalregion(Zhaoetal.,2019). As aresult,mostoftheunemployedpopulationinthelabor mar-ket triggered by water scarcity comes from food-related sectors.
Fig. 3 showsthat the job loss inagriculture accounted forabout
Fig. 6. Trade-offs and synergies of economic-social-environmental dimensions. Note: green plot is Y-Z axis, red X-Y axis, blue X-Z axis, and purple ball is X-Y-Z axis.
80%ofthetotaljoblossineachscenario,buttheshareof environ-mentalbenefit inagriculture is lessthan 10% of thetotal, which means that the relatively small extent of economic decline trig-gered by waterrationing in agriculture would lead to large fluc-tuations onthe agriculturallabor market anda smallergrey wa-ter footprint(Van Arendonk,2015). There arealso some opposite
Fig. 7. Scenario evaluation based on two MCDA theories. Note: Scenario ID are showed in Table 2.
cases with small change inlabor market but big change in grey waterfootprintinsomescenarios(e.g.B1->B2,T1->T2,H5->H6). These findingsindicate that whensocial andenvironmentalcosts areexcluded fromwaterconservationplanning,policymakersare morelikelytopromotesomewater-savingpoliciesattheexpense of joblossesandunexpected environmentalside effects,andthis resultcouldbemorepronouncedintheagriculturesectorbecause of its highwaterandjob intensity.A directpolicy implication of thisrecognitionisthatalthoughtheadjustmentinagriculturehas thelowestimpactontheeconomy,theresponsesofsocialand en-vironmentalaspectsshouldalsobeconsideredwhenpolicymakers proposewaterpolicyguidelines.
Fig.7showstheoverallscoresforeachscenariowithregardto their performance ineconomic,socialandenvironmental sustain-ability dimensionsforguidingfuture industrialtransitionpolicies inthecapitalregion.Itshouldbenotedthatwecouldachievethe goalofdecreasingthewaterstressleveltobealignedwiththe car-rying capacityoflocalwaterresources basedon arangeof plau-sible pathways.In other words, there isspace forthe stakehold-ers to join the discussion on the calibration of criterion weights andtochoosethemostsuitablescenariobasedontheirneeds.For example, Beijingisona pathwaywithan economic structure to-ward highlyadvancedservices andaway fromthe current water-intensive industries,thus theshare ofirrigationagriculture could be significantly reduced, even close to zero entirely. Tianjin at-tempts to become a research center formanufacturing and a pi-lotregionforfinancialreform,thussomemanufacturingwithhigh water intensities andheavy pollutions will be replaced by some hightechnology industries.Hebeiis treatedasanimportant eco-logical conservationregion,withmany naturalparkssurrounding the capital.Forthisgoal,to renewheavyindustries byupgrading productionprocessesandmodernizingtheindustrialbasewill be-come requiredin the nearfuture.To some extent,the visualized presentation of the links between policy-relevant requirements andthe industrialtransitions canbetter serve theknowledge co-productionprocessbetweenscientistsanddecision-makersin wa-terdemand-sidemanagement.
Ourfindings indicatethattrade-offsandsynergiesexist simul-taneously in terms of economic development, social well-being andenvironmentalprotection whenthe economic systemis con-strained by waterresource endowment. Waterpolicies aiming to save water in agriculture will have limited effects on economic developmentbut have negativeand significant influences on so-cial andenvironmental issues.Similarly, shrinking economic out-putintheSmeltingandpressing(processing)ofmetalsand Con-structionsectorswillhavesignificantenvironmentalvaluesbut re-ducedeconomic andemployment benefits. Thus, quantifying the interactions between water and other elements at sectoral scale is very important forwater policy planning.Many water studies havebeen confinedto sectorial“silo thinking”, without consider-ingthetrade-offsbetweenmultipledevelopmentindicators,which hasusuallyledtoinefficientregulatorydecisionsandspilloverand backfireeffectsleadingtoenvironmentalandsocialproblems else-where(Bizikovaetal.,2013).Ourresearchsuggeststhatthis well-constructed“nexus” approachhasgreatadvantagesonimpact as-sessments and scenario selection, and this approach could be a powerfulandintegrativetechniqueforassessingthetrade-offsand synergiesamongmultiplecriteriaunderresourceshortagesor sur-pluses,andforaddressingtheeconomic,social,environmental,and physicalcontextsofresourcesystemstoachievemorebalanced so-lutionsforpolicymakersandrelevantstakeholders.
The proposedframework can alsobe applied tomodeling po-tential impacts of supply chain disruptions fromexternal shocks such as thecurrent pandemic (see also forsimilar modeling ap-proachesGuanetal.(2020)andShanetal.(2020)). Theoutbreak ofCOVID-19,causedbySARD-CoV-2(severeacuterespiratory syn-drome coronavirus2) hasbecome the mostdisruptiveviral pub-lichealtheventsince the1918 influenza pandemica century ago (Gianietal., 2020). Somestrictlockdownmeasures liketravel re-strictions, quarantine, closing caterings and entertainmentplaces andsocialdistancingareenforcedbymanygovernmentswithhigh infection rate and confirmed cases to slow down the spread of COVID-19 (Guan etal., 2020). Thisnewly integratedtoolbox with mixedMRIO,scenarioanalysisandMCDAtheoryprovidesdecision
makers with theability to assessshort-term supply-chain effects ofdifferentcontainmentmeasurestorevealhowpandemic-related economiclosseswillbereallocatedalongsupply-chainsacross re-gions andindustrialsectorsandquantifyassociatedpotential eco-nomic, social andenvironmental impacts. Forinstance, travelling restrictionsandclosingpublicplaces(restaurants,pubs,sportsetc.) will bring about external economic shocks inseveral key service sectorsdirectly,thentheimpacts oftheseconstrainedsectorswill beenhancedthroughcomplexsupply-chainlinkages.Subsequently, secondary impacts likeunemployment, andchanges in air pollu-tion, water resources and biodiversity effects triggered by lock-down measures can be captured by this framework, and allows quantificationandcomparisonsbetweencostsandbenefitsof var-iousstrategies.
The successful achievement of the SDG agenda is a path-way to respond to the global sustainability challenges in ensur-ing economic prosperity, human well-being, and environmental protection, and the attainment of SDGs will greatly depend on whethersynergiescan beleveraged andtrade-offsare minimized (Pradhanetal.,2017).TheUNhasestablishedtheseSDGtargetson the global scale; however,assessing these targetsat theregional scale in a scientific way is more practical for local governments andpractitioners.Ouranalyticalframeworkcanbeonetoolto as-sesstrade-offsandsynergiesofSDGtargetsattheregionallevel.A couplingofthisregionaltoolwithnationalandglobalMRIO mod-els in a nested structure in future research would achieve more insightfulnationaloverviews,andhelpdetermineprioritiesin sup-portingSDGtargets.ThemixedMRIOmodelresultsshowthe ne-cessity ofconsidering the impacts of the disruption triggered by water constraints on the supply chain across region-sectors. Our results indicate that of the total economic loss triggered by the imposition ofwaterrationing in the capitalregion, 54% occurred in the capital region and45% occurredin the rest of China. We expect thatourintegratedapproachcanserveasastepping stone for further research into the cascadingrelationships among food security, energy consumption, and environmental restoration. In thesefields,therelativeintensitiesofpressures,trade-offs,and co-benefits will depend onthe scope of each analysisandthe indi-cators used to measure the outcomes (Obersteiner et al., 2016). We believe that thesimplicityand therich informationprovided by MRIOmodelshaveadvantagesincontributingtocoherentand comprehensivepolicyplanning,withdueattentionfocusedon eco-nomicprosperity,resourcesecurityandsocialstability.
5. Conclusion
This study proposed a novel approach that integrates supply constrained MRIOmodel withmulti-criteria assessment to quan-tifythetrade-offsandsynergiesofeconomic-social-environmental dimensions at the regional-sectoral level with an application to thecapitalregionofChina.Wedeveloped120industrialtransition scenarios includingnineindustrieswiththehighwater-intensities andwaterconsumptionundercurrentdevelopmentpathways.We calculated economic loss, job loss and environmental protection gainstriggeredbywaterconservationmeasuresundereachofthe 120scenarios.Then,weemployedMCDAtoselectoptimal scenar-ios.Thefollowingconclusionsweredrawn:
1. A tolerable water rationing in Agriculture, Food and tobacco processing, Electricity and heating power production, and Chemicalssectorswouldresultinthelowesteconomicandjob lossesaswell asthelargest environmentalbenefits tothe re-gion.
2. TheMCDAprocedurerecommends10referencescenarios with regard to their economic, social and environmental perfor-mance that can facilitate the design of future water
regula-tion and industrial transition policies. The visualized presen-tation of the links between policy-relevant requirements and the industrial transitions can better serve the knowledge co-productionprocess between scientists anddecision-makers in waterdemand-sidemanagement.
3. Thisnewly integrated toolbox allows toassess the short-term supply-chain effects of different crisis containment strategies toreveal how pandemic-related economic losseswill be real-locatedalongthesupply-chainsacrossregionsandsectorsand potentialotherimpacts,andallowsquantificationand compar-isonsbetweenthelossesandgainsofvariousmitigation strate-gies.
4. This integrated approach could be a powerful policy support toolforassessingtrade-offsandsynergiesamongmultiple cri-teriaunderresourceconstraintsandforevaluatingSDGsatthe regionalleveltodetermineprioritiesforlocalgovernmentsand practitioners.
Authorcontributions
J.L.,D.Z.,L.S.,O.V.,K.H andK.F.conceivedthecentralidea.D.Z. collectedthedata,performedthecalculations, andcreatedall fig-ures. D.Z.wrote the draft. Allauthors contributedto theanalysis anddevelopedthemanuscript.
Dataavailability
Detailsonthemethodologyanddataforestimatingwater con-sumptioninChinaaresummarizedinthesupplementary informa-tion,andanyotherdatasetsgeneratedduringthisstudyare avail-ableuponrequestfromtheauthors.
DeclarationofCompetingInterest
None.
Acknowledgments
Thisstudywasmainly supported by the NationalNatural Sci-enceFoundationof China (41625001)and AaltoUniversity. Addi-tionalsupportwasprovidedbytheStrategicPriorityResearch Pro-gram of Chinese Academy ofSciences (Grant no.XDA20060402); the National Natural Science Foundation of China (51711520317; 41571022); and the High-level Special Funding of the South-ern University of Science andTechnology (Grant No. G02296302, G02296402). The paper wasdeveloped within the framework of thePantaRheiResearchInitiative oftheInternationalAssociation ofHydrologicalSciencesandtheKeyLaboratoryofIntegrated Sur-faceWater-Groundwater PollutionControl,Southern University of ScienceandTechnology.Thepresentworkwaspartiallydeveloped withintheframeworkofthePantaRheiResearchInitiativebythe workinggroup “WaterScarcity Assessment:Methodologyand Ap-plication”.
Supplementarymaterials
Supplementary material associated with this article can be found,intheonlineversion,atdoi:10.1016/j.watres.2021.116986.
References
Afgan, N.H. , Carvalho, M.G. ,2008. Sustainability assessment of a hybrid energy sys- tem. Energy Policy 36, 2903–2910 .
Allan, J.A. , 1996. Water use and development in arid regions: Environment, eco- nomic development and water resource politics and policy. Rev. Eur. Commu- nity Int. Environ. Law 5 (2), 107–115 .
