利用机器学习和现代数据科学对司法改革成本和收益进行加权（英）

PublicDisclosureAuthorized PublicDisclosureAuthorized PolicyResearchWorkingPaper10449 WeightingJusticeReformCostsandBenefitsUsingMachineLearning andModernDataScience ChrisMahonyMatthewManningGabrielWong SocialSustainabilityandInclusionGlobalPracticeMay2023 PolicyResearchWorkingPaper10449 Abstract Cantheimpactofjusticeprocessesbeenhancedwiththeinclusionofaheterogeneouscomponentintoanexistingcost-benefitanalysisappthatdemonstrateshowbenefactorsandbeneficiariesareaffected?Suchacomponentrequires (i)movingbeyondthetraditionalcost-benefitconceptualframeworkofutilizingaverages,(ii)identificationofsocialgrouporpopulation-specificvariation,(iii)identificationofhowjusticeprocessesdifferacrossgroups/populations,(iv)distributionofcostsandbenefitsaccordingtotheidenti-fiedvariations,and(v)utilizationofempiricallyinformedstatisticaltechniquestogainnewinsightsfromdataandmaximizetheimpactforbeneficiaries.Thispaperoutlinesamethodforcapturingheterogeneity.Thepaperteststhemethodandthecost-benefitanalysisonlineappthatwasdevelopedusingprimarydatacollectedfromadevelop-mentalcrimepreventioninterventioninAustralia.The paperidentifieshowsubgroupsintheinterventiondis-playdifferentbehavioraladjustmentsacrossthereferenceperiod,revealingtheheterogeneousdistributionofcostsandbenefits.Finally,thepaperdiscussesthenextversionofthecost-benefitanalysisapp,whichincorporatesanartificialintelligence-drivencomponentthatreintegratesindividualcost-benefitanalysisprojectsusingmachinelearningandothermoderndatasciencetechniques.Thepaperarguesthattheappenhancescost-benefitanalysis,developmentoutcomes,andpolicymakingefficiencyforoptimalprioritizationofcriminaljusticeresources.Further,theappadvancesthepolicyaccessibilityofenhanced,socialgroup-specificdata,illuminatingoptimalpolicyorientationformoreinclusive,just,andresilientsocietaloutcomes—anapproachwithpotentialacrossbroaderpublicpolicy. ThispaperisaproductoftheSocialSustainabilityandInclusionGlobalPractice.ItispartofalargereffortbytheWorldBanktoprovideopenaccesstoitsresearchandmakeacontributiontodevelopmentpolicydiscussionsaroundtheworld.PolicyResearchWorkingPapersarealsopostedontheWebathttp://www.worldbank.org/prwp.Theauthorsmaybecontactedatcmahony@worldbank.org. ThePolicyResearchWorkingPaperSeriesdisseminatesthefindingsofworkinprogresstoencouragetheexchangeofideasaboutdevelopmentissues.Anobjectiveoftheseriesistogetthefindingsoutquickly,evenifthepresentationsarelessthanfullypolished.Thepaperscarrythenamesoftheauthorsandshouldbecitedaccordingly.Thefindings,interpretations,andconclusionsexpressedinthispaperareentirelythoseoftheauthors.TheydonotnecessarilyrepresenttheviewsoftheInternationalBankforReconstructionandDevelopment/WorldBankanditsaffiliatedorganizations,orthoseoftheExecutiveDirectorsoftheWorldBankorthegovernmentstheyrepresent. ProducedbytheResearchSupportTeam WeightingJusticeReformCostsandBenefitsUsingMachineLearningandModernDataScience ChrisMahony1*,MatthewManning2*,GabrielWong3* Keywords:justicereform;cost-benefitanalysis;machinelearning;datascience;justiceprocesses;heterogeneity *ThispaperexclusivelyreflectstheviewsoftheauthorsanddoesnotnecessarilyrepresentthoseoftheWorldBanknoritsBoardofDirectors.TheauthorsthankVaralakshmiVemuru,JoseCuesta,RohiniSrihari,JamesBrumby,BarryMaher,PatrickBarron,andAlyRahimforusefulfeedbackandsuggestions.Allremainingerrorsaresolelytheresponsibilityoftheauthors. 1WorldBankGroup;PeloriaP.B.C; 2DepartmentofSocialandBehavioralSciences,CityUniversityofHongKong; 3ANUCentreforSocialResearchandMethods,TheAustralianNationalUniversity Introduction ThejointUnitedNations-WorldBankflagshipreport,'PathwaysforPeace'(UnitedNationsandWorldBank,2018),citesgrievancessurroundingsocialgroup-specificexclusionfromaccesstojusticeandsecurityasoneoffourarenasofsocialcontestthatinformtheriskofviolence.1TheWorldBankhasdevelopedanalyticaltoolsforidentifyingthepoliticaleconomyandefficacyofgovernanceandjusticereformincludingtherelationshipbetweenjusticesectorpolicyandprogrammaticapproachesandtheWorldBank’stwingoalsofreducingextremepovertyanddrivingsharedprosperity.2 Toachievetheabovegoals,acomprehensiveeconomicframeworkisrequiredfortracinghowpublicpolicyprocessesortheirsubstantiveconsequencesvaryacrossindividualsandcommunities.Currentlyinthisspace,however,cost-benefitanalysis(CBA)reliesontheaveragetreatmenteffect(ATE),whichdoesnotadequatelyunpackthecontextualvariationsthatmoderateeconomicoutcomesacrossaffectedcommunitiesorsub-groupswithincommunities. Thispaperconsidersthepolicyquestionspecifictothejusticesector:CantheimpactofjusticeprocessesbeenhancedwiththeinclusionofaheterogenouscomponentintoanexistingCBAAPPthatdemonstrateshowbenefactorsandbeneficiariesareaffected?Thedevelopmentofsuchaframeworkrequires:(i)movingbeyondthetraditionalcost-benefitconceptualframeworkofutilizingaverages(i.e.ATE);(ii)identificationofvariationordifferencesbetweengroupsorpopulations(e.g.individualactorattributes,politicalstructures,institutions);(iii)identificationofhowjusticeprocessesdifferacrossgroups/populat