公共部門工資對腐敗的影響 -工資不平等問題 Effects of Public Sector Wages on Corruption - Wage Inequality Matters

EffectsofPublicSectorWages

onCorruption

WAGEINEQUALITYMATTERS

AsliDemirgü?-Kunt,MichaelLokshin,andVladimirKolchin

Abstract

Thepaperusesanewcountry-level,paneldatasettostudytheeffectofpublicsectorwagesoncorruption.Theresultsshowthatwageinequalityinthepublicsectorisanimportantdeterminantoftheeffectivenessofanti-corruptionpolicies.Increasingthewagesofpublicofficialscouldhelpreducecorruptionincountrieswithlowpublicsectorwageinequality.Incountrieswherepublicsectorwagesarehighlyunequal,however,raisingthewagesofgovernmentemployeescouldincreasecorruption.Theseresultsarerobusttoawiderangeofempiricalmodelspecifications,estimationmethods,anddistributionalassumptions.Combiningincreasesinpublicsectorwageswithpoliciesaffectingwagedistributioncouldhelppolicymakersdesigncost-effectiveprogramstoreducecorruptionintheircountries.

KEYWORDS

Corruption,

bureaucracy,

paneldataanalysis,

public-private

wagedifferential,

government

wagepolicy

JELCODES

J38,J45,O57,D73

WORKINGPAPER644?MAY2023

EffectsofPublicSectorWagesonCorruption:WageInequalityMatters

AsliDemirgü?-Kunt

CenterforGlobalDevelopment

MichaelLokshin

WorldBank

VladimirKolchin

WorldBank

Thispaper’sfindings,interpretations,andconclusionsareentirelythoseoftheauthorsanddonotnecessarilyrepresenttheviewsoftheWorldBank,itsExecutiveDirectors,orthecountriestheyrepresent.WethankToanDo,ZahidHasnain,StutiKhemali,MenahemPrywes,MartinRavallion,FrancescaRecanatini,RobyShenderowitsch,IvanTorre,andtheanonymousrefereefortheircomments.

AsliDemirgü?-Kunt,MichaelLokshin,andVladimirKolchin.2023.“EffectsofPublicSectorWagesonCorruption:WageInequalityMatters.”CGDWorkingPaper644.Washington,DC:CenterforGlobalDevelopment.

/publication/effects-public-sector-wages-corruption-wage-inequality-matters

CENTERFORGLOBALDEVELOPMENT

2055LStreet,NWFifthFloorWashington,DC20036

1AbbeyGardens

GreatCollegeStreet

London

SW1P3SE

CenterforGlobalDevelopment.2023.

TheCenterforGlobalDevelopmentworkstoreduceglobalpovertyandimprovelivesthroughinnovativeeconomicresearchthatdrivesbetterpolicyandpracticebytheworld’stopdecisionmakers.UseanddisseminationofthisWorkingPaperisencouraged;however,reproducedcopiesmaynotbeusedforcommercialpurposes.FurtherusageispermittedunderthetermsoftheCreativeCommonsLicense.

TheviewsexpressedinCGDWorkingPapersarethoseoftheauthorsandshouldnotbeattributedtotheboardofdirectors,fundersoftheCenterforGlobalDevelopment,ortheauthors’respectiveorganizations.

Contents

1.Introduction 1

2.Literaturereview 2

3.Dataandvariabledefinitions 4

3.1Maindependentvariables 4

3.2Othercontrolvariables 6

3.3Countryhistoricalandinstitutionalsetting 7

4.Empiricalspecification 8

5.Results 10

Explainingourresults 14

6.Sensitivityanalysis 15

7.Conclusions 17

References 18

Figuresandtables 23

Appendix 32

ListofFigures

1.Changesinthepublicsectorwagepremiumandcompressionratiofor

Argentina,Bolivia,andHungary,2000–2018 23

2.Effectofchangesinpublic-privatewagedifferentialoncorruption

fordifferentlevelsofwagecompressioninthepublicsector 24

3.Contourplotofpredictedlevelsofcorruptionasafunctionofpublic-private

wagedifferentialandwagecompressionratiointhepublicsector 25

ListofTables

1.Descriptivestatisticsfordependentandmainindependentvariables 26

2.FixedeffectestimationsofWGIcorruptionindex 27

3.Fixedeffectestimationsofthree(standardized)measuresofcorruption

andtwomeasuresofthepublic-privatewagedifferential 28

4.Estimationsofalternativemodelspecifications 29

5.Estimationsusingalternativeeconometrictechniques 30

6.Extremeboundsanalysisofcoefficientsonpublic-privatewagedifferential

andinteractiontermofwagecompressionratioandpublic–privatewage

differential 31

A1.Fixedeffectestimationsofthreemeasuresofcorruption

(original,unstandardizedindicators) 32

A2.Fixedeffectestimationofthemainspecificationofthreemeasuresofcorruption

andpublic-privatewagedifferentialfordifferentoccupations(comparedwithin

theformalsectoronly) 33

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:1

WAGEINEQUALITYMATTERS

1.Introduction

Anti-corruptionpoliciesinmanycountriesrelyonthenotionthatcorruptioniscausedbylowwagesinthepublicsector.Inattemptstocurtailcorruption,Argentina,Georgia,Ghana,Peru,Singapore,andothercountrieshaveimplementedpublicsectorreformstoincreasethewagesofgovernmentofficials.

Theevidenceontheeffectivenessofsuchinterventionsismixed.Somestudiesfindthathigherwagesinthepublicsectorwereassociatedwithlowercorruption(Klitgaad1997,VanRijckeghemandWeder2001,AnandKweon2017).Othersfindnosignificanteffect(Panizza2001,AdesandDiTella1997,Treisman20002007)orreverseeffect,withhighlevelsofcorruptionleadingtolowwagesinthepublicsector(Rose-AckermanandS?reide2012).

Differencesintheavailability,quality,andcomparabilityofdata,aswellasmethodologicalissuesrelatedtothepotentialeffectsofunobservablefactors,accountforthemixedresults(Treisman2007).Newlyavailablecross-countrydatafromtheWorldwideBureaucracyIndicatorsdatabaseallowustoaddressmanyoftheseproblemsandpresentnewevidenceontheeffectivenessofincreasingwagesasananti-corruptionmeasure.

Ourfindingssuggestthatthedistributionofwagesinthepublicsector(measuredasaratioofthe90thtothe10thwagepercentiles)couldbeanimportantdeterminantoftheeffectivenessofanti-corruptionpolicies.1Theheterogeneityintheimpactofhigherwagesoncorruptionwithrespecttowageinequalityinthepublicsectormaypartlyexplainwhysomeofthepreviousstudieshavefoundnosignificanteffectsofwagesoncorruption.Wefindthathigherwagesmayreducecorruptionincountrieswithrelativelycompressedwagesinthepublicsector.Incontrast,increasesinthewagesofpublicservantscanencouragecorruptionifpublicsectorwagesarehighlyunequal.Combiningtheincreasesinpublicsectorwageswithpoliciesaimedatreducingpublicsectorwageinequalitymightallowpolicymakerstodesigncost-effectiveprogramstolowercorruptionintheircountries.

Thelongitudinalstructureofourdataallowsustotacklearangeofeconometricissuesthatpreviousstudiescouldnotaddress.Wealsouseindicatorsobtainedfrommicro-levelsurveys.Mostcross-countrystudiesofcorruptionandwagesrelyonmacro-leveldatatoderivethepublic-privatewagepremium(e.g.,VanRijckeghemandWeder2001,AnandKweon2017,Treisman2000).Suchanapproachisassociatedwithpersistentmeasurementerrorsandfailstocontrolforage,gender,education,location,andotherindividualcharacteristicsinderivingthepublicwagepremium(Schiavo-Campoetal.1997,Leetal.2018).This“unadjusted”wagedifferentialcapturesthedifferencesbetweenthecharacteristicsofworkersbetweenthepublicandprivatesectorsandnotthedifferencesinreturnstothesecharacteristics.Ouranalysisreliesonan“adjustedpaypremium”

1Meyer-Sahlingetal.(2018)showthattheeffectofpaylevelsoncorruptionmaybecontextspecific,dependingonarangeofcharacteristicsofpaysystems,includingthepublicwagecompressionratio.Toourknowledge,ourpaperisthefirstempiricalworktoaddressthatissue.

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:2

WAGEINEQUALITYMATTERS

thatreflectsthewagedifferencesbetweencomparableworkersemployedinthetwosectors(Borjas2012).Ourresultsarerobusttoawiderangeofempiricalmodelspecifications,estimationmethods,anddistributionalassumptions.

Thefollowingsectionreviewstheliteratureontheeffectofpublicsectorwagesoncorruption.Section3describesthedataandmainvariables.Section4presentstheempiricalmodel.Section5presentsthemainresults.Section6addressestheendogeneityofthepublic-privatewagedifferential.Section7presentstherobustnesschecks,andSection8concludeswithpolicyimplications.

2.Literaturereview

Alargebodyofliteratureinvestigatestheeffectsofthecompensationofpublicsectoremployeesoncorruption.TheBeckerandStigler(1974)“shirkingmodel”predictsthatpublicofficialswillengageincorruptioniftheexpectedreturnsfromsuchactivitiesexceedtheirexpectedwageincomes.The“fairwage”hypothesisofAkerlofandYellen(1994)postulatesthatpublicofficialsengageincorruptionuntiltheirwagesrisetowhattheyperceiveasfairwages.Faircompensationofpublicemployeesmayleadsocietiestocondemncorruptionratherthanperceiveitasaninstilledculturalnorm(VanRijckeghemandWeder2001).Theempiricalevidenceontheeffectofpublicsectorwagesoncorruptionremainsinconclusive.

Theempiricalliteratureoncorruptionandwagesissparse,partlybecauseofthedifficultiesofcollectinggood-qualitydata(Gans-Morseetal.2018,OlkenandPande2012).Individualswhousepublicofficeforprivategainhavenoincentivestorevealinformationthatmaycompromisethem(e.g.,Ackerman2016).

FoltzandOpoku-Ageyemang(2015)investigatetheeffectofdoublingthesalariesofpoliceofficersonbribeextortionsfromtruckdriversinGhana.Theyfindthatthehikeinpolicesalaryincreasedtheamountandfrequencyofbribestruckdriverspaidtothepolice.Theauthorsconjecturethatthewagereformraisedthesocialstatusoftheofficersandchangedtheirreferenceofthe“fair”incomelevel,leadingtoupwardrevisionsoftheexpectedamountsofbribes.Mishraetal.(2008)lookattheeffectsofa1997payreforminIndiathatincreasedthewagesofcustomsofficials.Theyfindthereformhadnoimpactontariffevasion:officialskepttakingbribesatthesamerateafterreceivingpayincreases.Light(2014)arguesthatadrasticincreaseinthewagesofpoliceofficersaspartofthepolicesystem’sreformledtoareductionincorruptioninGeorgia.DiTellaandSchargrodsky(2003)studythenexusbetweenwagepremiumsandcorruptioninArgentina.Theyfindthatfrequentauditsreducecorruption.However,thehigherwagespaidtoprocurementofficersfailtoreducecorruptionwhentheprobabilityofdetectioniseitherloworveryhigh;largerwagepremiumscombinedwithintermediateauditinglevelsreducecorruption.

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:3

WAGEINEQUALITYMATTERS

Theevidenceontheimpactofpublicsectorwagecutsoncorruptionismoreconclusive.Borcanetal.(2014)investigatetheeffectofanunanticipated25percentwagereductioninpublicschoolsonthepassingratesofstandardizedexamsinRomania.Theyreportthatthepercentageofstudentswhopassedtheexamsinpublicschoolscomparedtoprivateschoolsincreased.Theauthorsattributethatdifferencetoincreasedcorruptionamongpublicschoolteachers.GorodnichenkoandPeter(2007)findthatpublicsectoremployeesinUkrainewhoareunderpaidrelativetotheircounterpartsintheprivatesectormaycompensateforthatdifferencebytakingbribes.Theynotethatthewagegapbetweenthepublicandprivatesectorswidensatthetopofthewagedistribution,suggestingthatdecompressingpublicsectorwagesmightcurbcorruption—similartowhatwefindinthispaper.

Whileprovidingvaluablecountry-levelevidence,single-countrystudiesofcorruptionmaysufferfromtheproblemofexternalvalidity.Theissuesofcorruptionaresomultidimensional,andtheeffectivenessofdifferentmeasurestofightcorruptiondependsonsomanyfactors(includinglegal,institutional,andculturalfactors,whichareoftenunobservabletoresearchers)thatitmaybedifficulttogeneralizetheexperienceofonecountrytoothers.

Severalstudiesanalyzetherelationshipbetweencorruptionandpublic-privatewagedifferentialsinacross-countrysetting.GoelandRich(1989)documentthattheincidenceofbriberyconvictionsofcivilservantsisinverselyrelatedtothepublic-privatewagepremiumacrossstatesintheUS.VanRijckeghemandWeder(2000)findanegativecorrelationbetweenthewagepremiumofpublicsectoremployeesandcorruptionindevelopingandlower-incomecountries.Theyestimatethatpayingpublicemployeestwicetheaverageprivate-sectorwageisassociatedwithadecreaseof0.5pointsonacorruptionmeasurethatrangesfrom0to6.

StudyingcountriesinLatinAmericaandtheCaribbean,Panizza(2001)findsnocorrelationbetweencorruptionandthepublic-privatewagedifferential.However,hereportsasignificantpositivecorrelationbetweencorruptionandthepublic-privatewagedifferentialforformalsectorworkerswithloweducation.Leetal.(2013)discoveranegativerelationshipbetweengovernmentwagesandcorruptioninalargesetofcountries.Theyfindthattheimpactofwagesoncorruptionisstrongerinlow-incomecountries.AnandKweon(2017)analyzeapanelof43countriesbetween1999and2008.Theyfindthatthepublicsectorwagepremiumhasamodesteffectonreducingcorruption.Theaveragenon–OECDcountrywouldneedtoincreasepublicsectorcompensationbyafactorof10toreduceitscorruptiontothelevelsoftheOECDcountries,accordingtotheirstudy.

Severalmacro-levelstudiesarguethatwagepremiumshaveverylimitedornoimpactoncurbingcorruption(AltandLassen2014,Dahlstr?metal.2012,RauchandEvans2000,Treisman2000).Dahlstr?metal.(2012)suggestthatcorruptionisaffectedbythemeritocraticrecruitmentofpublicworkersratherthanbytheirremunerationlevels.Treisman(2000)reportsaninsignificantcorrelationbetweenwagesandcorruptionindifferentspecificationsbasedonasmallsampleofcountries.

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:4

WAGEINEQUALITYMATTERS

AnandKweon(2017)istheonlystudyknowntousthataddressesomittedvariablebiasbyapplyingcountryfixedeffect(FE)estimation.Allotherstudiescitedaboverelyonestimationsthatexploitcross-countryvariationandthereforehaveagreaterriskofintroducingomittedvariablebias.Inaddition,allthecross-countrystudiesmentionedabove,exceptLeetal.(2013)andPanizza(2011),usemacro-leveldatatoimputetheaveragewagesofpublicandprivatesectorworkers.AsweargueintheIntroduction,wagedifferentialsestimatedfrommicro-levelsurveydataholdseveraladvantagesoverdifferentialsconstructedfrommacrostatistics.

3.Dataandvariabledefinitions

3.1Maindependentvariables

Weusethreemeasuresofcorruption.TheControlofCorruptionIndicatorcomesfromtheWorldwideGovernanceIndicators(WGI)databaseproducedbytheWorldBankannuallysince1996forover200countries(Kaufmannetal.2010).Theindicatorreflectsperceptionsofbothpettyandgrandformsofcorruptionandthe“capture”ofthestatebyelitesandprivateinterests.WerefertothismeasureasCorruption_WGI.

Since2007,theWorldEconomicForum(WEF)hasproducedtheEthicsandCorruptionIndicatorforover140countries(WorldEconomicForum2018).Itreflectsrespondents’perceptionsofwhethergovernmentspreventtheillegaldiversionofpublicfundsandhowfrequentlyinvestorsandcompaniesmakeunofficialpayments.WerefertothismeasureasCorruption_WEF.

TheCorruptionPerceptionIndex(CPI)comesfromTransparencyInternational(2020).Itisacompositeindexofcorruptioninthepublicsectorasperceivedbyexpertsandbusinesspeople.TIhasbeenproducingtheCPIannuallysince1996forover180countries.Achangeinthemethodologyin2012constrainsthetimeseriestotheperiodfrom2012to2018.WerefertothismeasureasCorruption_TI.2Westandardizeallcorruptionindicatorstobeofmean0andstandarddeviation1.Tosimplifytheinterpretationandcomparisonofourresults,weinvertthescaleforWGIandWEFcorruptionindicatorssothatlargervaluesofstandardizedindicatorscorrespondtomorecorruption.Weshowthatthemainresultsofourpaperusingtheoriginal,unstandardized,valuesofthe

2Weconsideredusingthreeotherindicatorsofcorruption:theAbsenceofCorruptioncomponentoftheWorldJusticeProject’s(WJP)RuleofLawIndex(WorldJusticeProject2020),theAnti-CorruptionPolicycomponentof

theBertelsmannTransformationIndex(BTI)(BertelsmannStiftung2020),andtheCorruptioncomponentoftheInternationalCountryRiskGuide(ICRG)(PRSGroup2020).Thefirsttwoindicatorssignificantlyreducethenumberofcountriesinthesample.TheWJPdataarenotcomparablebefore2015,whichresultsinfewerthan80observationsfor17countries.TheBTIincludesonlydevelopingcountriesandtransitioneconomies,limitingthenumberofcountriesinouranalysisto19.WedecidednottousetheICRGindicatorfollowingKnack(2007)andTreisman(2007),whoadviseagainstusingthiscross-countrycorruptionindicatorforlongitudinalanalysis.

WAGEINEQUALITYMATTERS

corruptionindicatorsarequalitativelysimilartotheresultsbasedonthestandardizeddependentvariables(Table2AintheAppendix).3

Table1summarizesthestatisticsforthemainvariablesusedinouranalysis.TheWGIcorruptionindexranksParaguayandtheRussianFederationasthemostcorruptcountriesinoursampleandFinlandastheleastcorrupt.FinlandalsohasthelowestscoreontheTIcorruptionindex,onwhichHondurasandGreecerankasthemostcorrupt.4

MostofourcontrolvariablescomefromtheWorldwideBureaucracyIndicators(WWBI)dataset(WorldBank2020b).Themainfeatureofthesedataisthatitderivescountry-levelindicatorsfrommicro-levellaborforcesurveys.Differencesbetweenthepublicandprivatesectorwagesderivedfrommacrodataignorethedifferencesinworkers’humancapitalcharacteristicsthatleadtopotentialbiaseswhencomparedwithdifferencesobtainedfromthemicro-levelsurveys(Leetal.2018).5Anydistributionalstatistics,suchaswagecompressionratios,cannotbederivedfrommacro-leveldataatall.

TheWWBIdatasetisapanelof132countriescoveringtheperiod2000–19.Forty-fourcountriesinthedatasethaveatleast4panelobservations,41countrieshaveatleast6,36haveatleast8,and33haveatleast10.Thesamplehas454or507country-yearobservationsdependingonthemeasureofthepublic-privatewagedifferentialweuse.

Thepublicsectorwagepremiumisestimatedasacoefficientonthepublicsectoremploymentdummyinthestandardlog-earningsregressionsforeachcountry(Mincer1974).6Thesamplesizesfortheseestimationsrangefrom6,799observationsforRussiato1.8millionforBrazil.TheWWBIdatasetcontainstwostandardmeasuresofthepublicsectorwagepremium(WPi,t).Thefirstisestimatedonasampleofpublicsectorworkersandtheircounterpartsintheformalprivatesector.

3Severalstudies(e.g.,Knack2007)questionthevalidityofusingtheWGIandTIcorruptionindicatorsforlongitudinalanalysis.Changesindatasourcesandinthemethodologyofconstructingindicatorsareamongthemainissuesaffectingtheresultsoflongitudinalestimates.WeareawareofthesecritiquesandaddresstheminSection6byreplicatingourresultsonpanelsofdifferentlengthsandusinglaggedindependentvariables.Additionally,theWGIindicatorisconstructedbasedonabroadersetofdatasources.Knack(2007),Rohwer(2009),andChabova(2017)discussdifferencesandmethodologicalissuesrelatedtotheconstructionofthecorruptionindicators.Inlightofthesestudies,theWGIindicatoristhemostreliableinassessingthelevelsofcorruptioninacountry.

4RussiaandParaguayarenotincludedintheTIsamplebecausebothcountrieshavefewerthanfivelongitudinalobservationsbetween2012and2018.

5Forexample,ifthepublicsectorworkersaremoreexperiencedandeducatedcomparedwiththeircounterpartsintheprivatesector,themacro-levelanalysiswouldoverestimatethepublicwagepremium.Thisproblemislargelyalleviatedwhendifferencesarederivedfrommicrolevelsurveysthatcancontrolforworkers’characteristics.

6Thepublicsectorincludesthecentralgovernment,nongovernmentalorganizations,thearmedforces,andstate-ownedcompanies.Theprivatesectoristhepartoftheeconomythatisbothrunforprivateprofitandnotcontrolledbythestate(WorldBank2020b).Thepublic–privatewagedifferentialisestimatedonthesampleofemployeesineachcountry.Formally,theempiricalspecificationforthisestimationisln(wagei)=a+bPublici+冗Xi+ci,where

Publiciequals1ifapersonisemployedinthepublicsectorandXiisthesetofcontrols,whichincludeage,agesquared,gender,education,andlocationofaworker.Toderivethepublic-privatewagedifferential,theestimatedcoefficientβisdeloggedandreducedby1sothatthewagedifferential=(exp()?1).Theresultingwagedifferentialsarenegativeifpublicsectorwagesarelowerthantheprivatesectorwagesandpositiveotherwise(Gindlingetal.2020).

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:5

EFFECTSOFPUBLICSECTORWAGESONCORRUPTION:6

WAGEINEQUALITYMATTERS

Thesecondisestimatedonasampleofemployeesofboththeformalandinformalpartsoftheprivatesector.Ourbaselinespecificationusesthewagedifferentialbetweenthepublicandformalprivatesectorworkers,aspublicemployeesaremorelikelytocomparetheirwageswithwagesintheformalprivatesector(GoelandRich1989).

TheWWBIdatasetalsoincludestwopublicsectorwagepremiumindicatorsderivedfordifferentoccupations.Thefirstcomparesthewagesofseniorpublicofficialsinthepublicsectorwithemployeesinrelatedoccupationsintheformalprivatesector.Thesecondcomparesthewagesofallprofessionalswithemployeesincorrespondingoccupationsintheformalprivatesector.Weusetheseindicatorstovalidateourmainresults.

Inoursample,onaverage,publicsectorworkersearn5.6percentmorethantheircomparatorsintheformalprivatesector.Thepublicsectorpremiumincreasesto15.1percentwheninformalsectoremployeesareaccountedfor,amagnitudesimilartothatofGindlingetal.(2020).Usingtheformalprivatesectorforcomparison,Peru(?34.2percent)andtheDominicanRepublic(?30.1percent)havethelowestdifferentials,andEcuador(50.9percent)andCyprus(48.2percent)havethelargest.Whenallworkersintheprivatesector(formalandinformal)areusedforcomparison,Russia(?29.0percent)andtheDominicanRepublic(?16.4percent)havethelowestpublic-privatewagedifferentials,andCostaRica(74.0percent)andPakistan(69.2percent)havethelargest.

Ourmeasureofwagedispersion—thewagecompressionratio(WCi,t)—isdefinedasaratioofthe90thtothe10thpercentilesofpublicsectoremployees’weeklywages(e.g.,Heyman2008,Almeida-SantosandMumford2005,Brunello2001)7.TheSlovakRepublic(2.4)andCroatia(2.6)havethelowestwagecompressionratiosinoursample.ThepublicsectorwagedistributionismostunequalintheRussianFederation(10.3)andBrazil(9.5).8

3.2Othercontrolvariables

Weuseseveralcontrolvariablesinourestimations.TheshareofpublicworkerswithtertiaryeducationcomesfromtheWWBIdataset.ThesmallestshareisinUruguay(28.3percent),andthelargestsharesareinLithuania(83.9percent)andIreland(78.0percent).

TheInternationalCountryRiskGuide(ICRG)datasetprovidestheindexofthequalityofthebureaucracy(PRSGroup2020).Itgaugeshowwellbureaucraticinstitutionscandeliverpublicservicesunderpoliticalpressure,especiallywhengovernmentschange.Theconjectureisthatastrongprofessionalbureaucraticbodycancounterattemptsbynewlyelectedpoliticianstoseekeconomicrents.Dahlstr?metal.(2012)showthevitalroleprofessionalbureaucratsmayplayin

7Onecanuse