fiscal rules and compliance expectations – evidence for the...
Post on 31-May-2020
5 Views
Preview:
TRANSCRIPT
Dis cus si on Paper No. 14-034
Fiscal Rules and Compliance Expectations –
Evidence for the German Debt Brake Friedrich Heinemann, Eckhard Janeba, Christoph Schröder, and Frank Streif
Dis cus si on Paper No. 14-034
Fiscal Rules and Compliance Expectations –
Evidence for the German Debt Brake Friedrich Heinemann, Eckhard Janeba, Christoph Schröder, and Frank Streif
Download this ZEW Discussion Paper from our ftp server:
http://ftp.zew.de/pub/zew-docs/dp/dp14034.pdf
Die Dis cus si on Pape rs die nen einer mög lichst schnel len Ver brei tung von neue ren For schungs arbei ten des ZEW. Die Bei trä ge lie gen in allei ni ger Ver ant wor tung
der Auto ren und stel len nicht not wen di ger wei se die Mei nung des ZEW dar.
Dis cus si on Papers are inten ded to make results of ZEW research prompt ly avai la ble to other eco no mists in order to encou ra ge dis cus si on and sug gesti ons for revi si ons. The aut hors are sole ly
respon si ble for the con tents which do not neces sa ri ly repre sent the opi ni on of the ZEW.
FiscalRulesandComplianceExpectations‐EvidencefortheGermanDebtBrake
FriedrichHeinemann(ZEWMannheimandUniversityofHeidelberg)
EckhardJaneba*(UniversityofMannheim,CESifoandZEW)
ChristophSchröder(ZEWMannheim)
FrankStreif(ZEWMannheim)
May2014
Abstract: Fiscal ruleshavebecomepopular to limit deficits andhighdebtburdens inindustrializedcountries.Agrowingliteratureexaminestheirimpactbasedonaggregatefiscalperformance.Sofar,noevidenceexistsonhowfiscalrulesinfluencedeficitexpec‐tationsof fiscalpolicymakers. In thecontextof theGermandebtbrake,westudy thisexpectationdimension.Inafirststep,weintroduceasimpledynamicmodelinanenvi‐ronmentcharacterizedbythelaggedimplementationofanewrule.Laggedimplementa‐tioncharacterizesthesetupoftheGermandebtbrakeandraisescredibilityissues.Inasecondstep,weanalyzeauniquesurveyofmembersofall16Germanstateparliamentsandshowthatthedebtbrake’scredibilityisfarfromperfect.Theheterogeneityofcom‐pliance expectations in the survey closely corresponds to our theoretical predictionsregarding states’ initial fiscal conditions, specific state fiscal rules and bailout percep‐tions.Inaddition,thereisarobustasymmetryincomplianceexpectationsbetweenin‐sidersandoutsiders(bothforin‐statevsout‐of‐statepoliticiansandthegovernmentvsopposition dimension),whichwe attribute to overconfidence rather than noisy infor‐mation.Theseresultssuggestthatnationalfiscalrulescanbestrengthenedthroughno‐bailoutrules,sustainableinitialfiscalconditionsandcomplementarysub‐nationalrules.JELClassificationCode:H6,H7Keywords:BudgetDeficits,DebtBrake,Credibility,Survey,FiscalRules Acknowledgement: The authors gratefully acknowledge support from the Collaborative Research Center (SFB) 884 “Political Economy of Reforms”, funded by the German Research Foundation (DFG).*Correspondingauthor:
EckhardJaneba,DepartmentofEconomics,UniversityofMannheim,L7,3‐5,68131Mannheim,Germany,++49‐621‐1811795,janeba@uni‐mannheim.de.
1
1.Introduction
Constitutional fiscal rules such as a balanced budget requirement have been used for
decadesinfederalcountriessuchasSwitzerlandandtheUSstatestolimitdeficitsand
debtsofsub‐nationaljurisdictions(forasurveyofcurrentfiscalrulesseeIMF,2012).On
thenational level, theeuroareadebtcrisishas triggeredawaveofnewstatutoryand
constitutionalbudgetconstraintsinordertoboostpublicborrowerreputation.TheFis‐
calCompact,acceptedbyallEUcountriesexcepttheUKandtheCzechRepublicin2012,
hasbeenanothermilestoneforthespreadofnumericalfiscalconstraints.Thesignatory
countriescommittotheintroductionofnationaldebtbrakeswithwell‐definednumeri‐
calcontents(i.e.governmentsarerequiredtolimitstructuraldeficitstoamaximumof
0.5%ofGDP,andtolowerdebtlevelssystematicallywhenexceeding60%ofGDP,Euro‐
peanCouncil,2011).
Agrowing literatureexaminesthe impactofnumerical fiscalrulesbasedonaggregate
fiscalperformanceindifferentregionalcontexts.Thestandardapproachistheestima‐
tionof cross‐sectionorpanelmodels for the selected jurisdictionsand theirdeficitor
debtperformance (for theUS seeEichengreenandBayoumi,1994,Poterba,1996; for
Europe seeDebrun,2000,LagonaandPadovano,2007,Debrunet al., 2008; forOECD
countriesseeDahanandStrawczynski,2010;andforSwisscantonsandmunicipalities
see Feld andKirchgässner, 2008;Krogstrup andWälti, 2008).A shortcomingof these
highlyaggregatedapproachesisthattheydonotrevealhowfiscalrulesimpactonthe
beliefs of fiscal decisionmakers regarding the credibility of the fiscal rule and hence
theirexpectationsforcompliance.Thepresentpaperaimstofillthisholebyexamining
theoreticallyandempiricallytheintermediatestepbetweenfiscalrulesontheonehand
anddecisionmakers’expectationsonfuturefiscaloutcomesontheotherhand.
Adefiningcharacteristicofaneffectiveandcredibleruleisthatitanchorsexpectations
consistent with the rule’s constraints. This logic has long been the key for assessing
monetaryrules. In themonetarycontext, incentivestogeneratesurprise inflationmay
undermine the credibility of an inflation rule (Kydland andPrescott, 1977;Barro and
Gordon,1983).Hence,amonetaryrule’seffectivenesscanbeassessedbyanalyzingits
impactoninflationaryexpectations.Forfiscalpolicy,ananalogyappliesforadeficitrule
anditsimpactoncomplianceexpectations.Duetothepoliticalcostsoffiscalconsolida‐
tion, politiciansmay face incentives not to complywith the deficit rule in the future.
2
Whetherandtowhatextentthenewruleisseenasabindingconstraintshouldbede‐
tectable fromcompliance expectations.This shouldhold inparticular for the expecta‐
tionsofthosepoliticalactorswhotaketherelevantbudgetarydecisions.Therefore,the
impactoffiscalrulesonpoliticians’expectationsoffersanaturalwaytoassessthecred‐
ibilityandeffectivenessofanewrule‐basedfiscalregime.
Althoughtheunderlying idea isofobviousandfundamental importance,ourcontribu‐
tionisthefirsttostudythroughasurveyofpoliticianstowhatextentpolicymakersex‐
pectcompliancewithanewfiscalrule.Thus,itservestofillastrikinggapinthelitera‐
tureonfiscalrules’effectiveness.Theapproachhasafurtherstrength:Whilestudieson
the linkbetweenrulesandobservable fiscalperformanceareonlyapplicableonanex
post‐basis(i.e.aftermanyyearsofexperiencewithanexistingrule)ourmethodcanbe
employedexante(i.e.oncearulehascomeintoexistencebutperformancedataarenot
yetavailable).Furthermore,itopenstheblackboxofaggregationand,instead,looksinto
theimpactofafiscalruleonthe(heterogeneous)expectationsofthoseindividualpoliti‐
cianswhoactuallytakethebudgetarydecisions.
TheinstitutionalcontextofouranalysisistheGermandebtbrake.Weexploreexpecta‐
tion formation for themembersof allGerman stateparliaments regardinganexisting
fiscal rule that becomes binding only several years from now. The case of Germany’s
debtbrakeisofinterestfortheunderstandingoffiscalrulesmoregenerallyandbeyond
Germany:First,theGermangovernmenthasbeenamajoradvocateforestablishingthe
FiscalCompactinEurope.Infact,inmanydimensionstheprovisionsoftheFiscalCom‐
pactare similar to thatofGermany’sdebtbrake.Therefore,abetterunderstandingof
theGermandebtbrakewillalsobehelpfulforassessingtheCompact’sconsequencesfor
other EU countries. Second, the German debt brake is characterized by lagged imple‐
mentationsinceitsbindingconstraintsarephasedinoveralongerperiod(forthecen‐
tralleveluntiltheyear2016andforthestatelevelin2020).Laggedimplementationisa
frequentstrategytorealizefarreachinginstitutionalreformssinceithelpstoovercome
reformresistance(seeBuchanan,1994,forageneraldiscussion).Atthesametime,the
transition process raises substantial credibility questions and the German debt brake
exampleofferstheopportunitytobetterunderstandthegeneralconditionsunderwhich
laggedimplementationcanneverthelessbecredible.
Ouranalysisofexpectationformationcomprisesatheoreticalandanempiricalcontribu‐
tion.Firstwedevelopatheoreticalmodelwiththreeperiods(labelled0,1,2)describing
3
thedynamic fiscal decision situation in an environment characterizedbyphasing in a
zerodeficit rule.Decisionsondeficits aredynamicbynatureand imply trade‐offsbe‐
tween instantand futurepoliticalcosts fromfiscalconsolidation.Themodel’skey fea‐
tureistheexistenceofafiscalrulewhichtakeseffectonlyinthefuture(period2).Afis‐
calshockinthenearfuture(period1)makescompliancewiththefiscalruleuncertain
whenthefiscalruleisnotcredible.Inperiod1,thegovernmenttradesoffthebenefits
andcostsofadheringtothefiscalrule.Complianceismorelikelytheloweristheinitial
deficit inperiod0 , thehigher isthegovernment’scompetence insmoothingthe fiscal
shock,thelowerarebailoutexpectations,thetighterisafiscalruleatthestatelevelin
period1,andthehigher thedeficit reduction inperiod0.Themodelalsoallowsus to
consider noisy information about a government’s ability and overconfidencewith re‐
specttocompliancewiththefiscalrule,whichcanexplainheterogeneousbeliefsofpoli‐
ticians.ThemodelthuscaptureskeyfeaturesoftheGermandebtruleaswellassimilar
institutionalinnovationselsewhereandguidestheempiricalanalysis.
Inasecondstep,wetestthesemodelpredictionsonthedriversofcomplianceexpecta‐
tions based on a unique survey ofmembers of all 16 German state parliaments,who
havebeencontactedwithaquestionnairerelatingtothenewdebtbrake.Inthesurvey
weelicitedresponsesforthepoliticians’expectationsontheownstatecomplyingwith
the new rule by the year 2020, on other states’ compliance, and on the likelihood of
sanctionsorbailout if a state violates thenewrule in2020. Since the surveywasnot
anonymous, individualcharacteristics (suchaseducation,partymembership,etc.)and
statecharacteristics(suchascurrentfiscalpositionandfutureneedforfiscalconsolida‐
tion)canbeusedtosystematicallystudythedeterminantsofcomplianceexpectations.
Thesurveyshowsthatthedebtbrake’scredibilityisfarfromperfect.Theheterogeneity
ofcomplianceexpectationsinthesurveycloselycorrespondstoourtheoreticalpredic‐
tions. States’ initial fiscal conditions, specific state fiscal rules andbailout perceptions
matter. In addition, there is a robust asymmetry in compliance expectations between
insidersandoutsiders(bothforin‐stateversusout‐of‐statepoliticiansandthegovern‐
ment versus opposition dimension). Insiders tend to be significantly more optimistic
thanoutsidersregardingthelikelihoodofstate’scompliance.Basedontheguidanceof
our theoretical model we diagnose overconfidence of insiders (and not noisy infor‐
mation)asdrivingthisasymmetry.Thesedetailedinsightsimproveourunderstanding
onhowthecredibilityofanewnationalfiscalrulecanbestrengthenedingeneral.Our
4
resultspoint to the importanceofno‐bailout rules, sustainable initial fiscal conditions
andcomplementarysub‐nationalrules.
Ourpaperisrelatedtovariousotherliteratures.Afewrecentpapersanalyzetheoreti‐
cally the role of fiscal rules in a political economy framework, such as Azzimonti,
BattagliniandCoate(2008).Janeba(2012)considerstheroleofdelayinmakingaGer‐
man type debt brake bindingwhen the fiscal rule itself is credible. The incentives of
bailoutsinafederalcontextareconsideredbyGoodspeed(2002).Kirchgässner(2002)
andVoigtandBlume(2011)examineempiricallytheeffectsoffiscalconstraintsonfiscal
outcomes.Expectationeffectsoffiscalruleswithrespecttobondmarketinvestorsplaya
roleforstudieswhichlookintotheimpactoffiscalrulesonriskpremiaofgovernment
bonds (Heinemann,Osterloh andKalb, 2014; Iara andWolff, 2014). Surveys of politi‐
cianshavebeenusedinrecentresearchbytwoofthepresentauthors.Heinemannand
Janeba(2011)useasurveyofmembersofGermany’snationalparliamenttostudyideo‐
logicalbiasintaxpolicy.JanebaandOsterloh(2013)useasurveyofmayorsintheGer‐
manstateofBaden‐Württemberg toempiricallymotivate thespatial structureof local
taxcompetitioninatheoreticaltaxcompetitionmodel.HeinemannandOsterloh(2013)
surveymembersoftheEuropeanParliamentregardingtheintroductionofaminimum
taxforcompaniesintheEUinordertodisentangleideologicalandnationalpreferences
ofpoliticians.
Therestofthepaperisorganizedasfollows.Section2setsupthetheoreticalmodeland
derivescomparativestaticsforthelikelihoodofcompliancewiththedebtbrake.Section
3describesouroriginalsurveyandprovidesbackgroundinformationonGermany’spo‐
litical and fiscal systemand thedebtbrake.Ourmain findings arepresented anddis‐
cussedinsection4.Finally,section5concludes.
2.AModelofFiscalRuleCompliance
Wemodelthedynamicfiscaldecisionofanincumbentgovernmenttoreduceitsdeficit
inordertomeetthetargetofafiscalrulebecomingeffectiveonlyinthedistantfuture.
Politicalcostsofdeficitreductionaremodeledinareducedforminordertofocusonthe
likelihoodofcompliancewiththefiscalrule.Lackofgovernmentcommitmentanddefi‐
citshocksmakecompliancenon‐trivialanduncertain.Specifically,weassumethat the
economylasts for threeperiods, 0,1,2.Themainvariableof interest is thegovern‐
mentdeficit .Theinitialdeficitisgivenby 0andisexogenousfromtheviewpoint
5
oftheincumbentgovernmentinperiod0.Thefiscalrule(i.e.,thedebtbrake)requires
thegovernmenttorunabalancedbudgetinperiod2.Ifthistargetismet(thatis,
0), thegovernmentobtainspayoff (c forcompliance).Otherwise thegovernment is
noncompliant andobtains payoff .Wedefine as the gross gain from
compliance.Thisdynamicset‐upcorrespondstothelaggedimplementationoftheGer‐
mandebtbrake.Periods0and1representthetime inwhichthedebtbrakehasbeen
implementedbutnotyettakeneffect.Period2,fromtheperspectiveofGermanstates,
wouldbetheyear2020inwhichthezerodeficitlimitfortheGermanstatesentersinto
force.
Thegovernmentcanreducethedeficitintwostepstowardthegoalbyreducingthedef‐
icitinperiods0and1bytheamounts 0and 0,respectively.Wemodeldeficit
reductioninareducedformwithoutspecifyingthenatureofthefiscaladjustment(i.e.,
taxincreasesand/orexpenditurecuts).Deficitreductioniscostlyforthegovernmentin
theperiodwhenittakesplacebecauseapprovalratingsofthegovernmentorreelection
chancesareharmed.Wefocusontheconcurrentcosteventhoughthecostofpermanent
deficitreductionmayspillovertofutureperiods.Wethusimplicitlyassumethatvoters
andpoliticianscaremostlyaboutthechangeofthedeficit,ratherthanitslevel.Thecost
functionforpermanentlyreducingthedeficitisc(r)intheperiodwhentheadjustment
ismade,andhasthepropertiesc’≥0,c’’>0,c(0)=0,andc’(0)=0.Strictconvexityim‐
pliesthatspreadingagivendeficitreductionovertimeisefficient,allelsebeingequal.
Thedeficitinperiod1isafunctionoftheinitialdeficitd0minusthereductionr0under‐
takeninperiod0.Thedeficitd1isstochasticduetoashockinfluencingthedeficitinpe‐
riod1.Theshockislabeledsandisdrawnfromtheuniformdistributionwithsupport
[0,S],whereS>0.Theprobabilitydensityfunctionisthus1/S.Therealizeddeficitshock
istheproductofsandanexogenousgovernmentcompetencemeasureq>0,whichre‐
flectstheabilityofthegovernmenttomoderateshocks.Lowerlevelsofqreflecthigher
ability.Whenputtingtheseelementstogethertheactualdeficitinperiod1is
. (1)
Inperiod1thegovernmentreducesthedeficitfurtherbychoosingr1sothat
. (2)
6
Byassumptionnoshockisassumedtotakeplaceinperiod2.Thegovernmentpayoffis
givenby
, (3)
where whenthegovernmentiscompliantinperiod2,thatisd2≤0,and
whennot.Let 1bethediscountfactor.1
2.1CredibleFiscalRule
Westartwithabenchmarksituationinwhichthefiscalrule 0iscredibleandthe
governmentmustcomplywith it regardlessof therealizationof theshock inperiod1
Therefore deficit reduction in period 1 is . The expected utility from
complianceisthen
, (4)
wherewemadeuseof theassumption that shocksareuniformlydistributedover the
interval[0,S].Thecostsofcompliancecomefromthecostofdeficitreductioninthefirst
periodplusthediscounted,probabilityweightedcostinperiod1,whichdependonthe
initialdeficitd0,period0deficitreductionr0,andthemagnitudeoftherealizedshockqs.
Thegovernmentinfluencestheexpectedutilitybychoosingr0,whichaffectsthedistri‐
butionofdeficitreductioncostsovertime.Theoptimalfirstperioddeficitreductionis
foundbymaximizing (4)with respect to r0.Theoptimum is implicitly givenby the
condition
. (5)
Strictconvexityofthecostfunctionensuresthatthesecondorderconditionholds.The
righthandsideof(5)representsthemarginalcostofincreasingdeficitreductioninpe‐
riod0.Thelefthandsidecapturesthemarginalbenefitofdoingso.Anincreaseinperiod
0deficitreductiondecreasestherangeoffeasibledeficitsinperiod1,whichonnetsaves
cost .
Insertingtheoptimalvalue into(4)givesthemaximalutilityfromcompliancewitha
credibledeficitruleandisdenoted . Weliketonotethatthereisnoguarantee
1 We could discount utility in period 2 by instead of . Doing so would simply rescale the utility level v, without affecting results. We omit the complication in order to save on notation.
7
thatagovernmentisbetteroffcomparedtonotcomplyingwiththefiscalrule(thelatter
isbyassumptionnotachoice,though).
2.2Lackofcommitment
Incontrasttotheprevioussection,wenowassumethatthecompliancedecisionisnot
forcedbyacrediblerule.Thecostofcomplianceinperiod1maybecomehighifthelevel
ofdeficitreductioninperiod0islowand/ortherealizationofthebudgetshockinperi‐
od1 isbad. Insuchasituation,agovernmentmayfinditattractivetonotcomply.We
analyze the conditions underwhich it is in the government’s interest to (not) comply
withthefiscaltarget,andifso,howthedeficitreductionisdistributedovertime.Forthe
timebeingwefocusonthepoliticaldecisionmakerandherinterestincompliance.Later
we consider how other individuals (such as opposition politicians or observers from
outsideofstate)assessthelikelihoodofcompliance.
Thestochasticnatureofthegovernmentdeficitinperiod1makesituncertainwhether
complianceoccurs.Animportantvariableinoursubsequentanalysisistheprobability
ofcompliancep.Weareinterestedintherelationshipbetweenpandexogenousparam‐
etersofthemodel,suchastheinitialdeficitd0,thegrossgainfromcomplianceu,gov‐
ernmentcompetenceq,possiblebailoutexpectations, aswell asadditional fiscal rules
restrictingthemaximumdeficitlevelinperiod1(priortotheexistingfiscalruleinperi‐
od2).Thelackofcommitmentrequiresthatwesolvethemodelbybackwardinduction.
Note that thegrossgainumustbepositive forcompliancewith the fiscal rule to take
placewithpositiveprobabilitybecausedeficitreductioniscostly.
Period1.
Thebinarypayoffstructure( , )combinedwithcostlydeficitreductionimpliesthat
wecanreducethegovernmentchoicesettotwooptions:Eithernotreachingthedeficit
goalinperiod2(andthereforenotspendinganyeffort,r1=0)orjustreachingthegoal
(withtheneedtosetr1=d1).Thelatterdominatestheformerif
, (6)
that is, the cost of reducing the deficit to zero is not higher than the gross gain from
compliance.Sincethecostofdeficitreductionc(r) isamonotonefunctionofr,wecan
invert(6)whenitholdswithequality,anddefineacritical leveloftheperiod1deficit
8
forcompliancetooccur,namely, ∗ .Ford1lessthanorequaltod1*,thegov‐
ernmentwill choose tobe compliant, otherwisenot.Using (1), the threshold levelde‐
finesimplicitlyamaximumlevelofthedeficitshocks,calleds*,thatisconsistentwithd2
=0.Thecriticallevelisgivenby
∗∗
. (7)
Insteadof statinggovernment compliance in termsofperiod1deficit (d1*), condition
(7)allowsustorestatethesamedecisionintermsoftherealizedvalueoftheshocks:
Fors≤s*thegovernmentwillbecompliant,otherwisenot.Thethreshold levels* isa
positivefunctionoftheadditionalgainfromcomplianceandofthedeficitreductionin
period 0, but depends negatively on initial deficit d0 and the inverse of government
competenceq(thelatteronlyunderappropriateassumptionsmadefurtherbelow).We
thuswrites*=s(r0;u,d0,q).Notethatr0isexogenousfromtheviewpointofperiod1,but
endogenousexante(unliketheotherthreevariables)anddeterminedinperiod0.
Givenauniformprobabilitydensityfunctionforswenowuse(7)tointroducetheprob‐
abilityofcompliancewiththefiscalrule
∗
. (8)
Theprobabilitypisthekeyobjectforourfurtheranalysisandliesbetween0and1un‐
dersuitableassumptionsonthesizeofd0andS.2Itdependson(r0;u,d0,q,S).Noteinpar‐
ticularthat
0, (9)
thatis,pincreases(decreases)withthelevelofperiod0deficitreduction(initialdeficit)
andthechangeisgivenbythecompetenceweightedprobabilitydensityofthevariable
s.Moredeficitreductioninperiod1makescompliancewiththefiscalrulemorelikely.
2 First, we assume that , which is sufficient to make s* in (7) nonnegative (because we assume r0 ≥ 0). The condition holds, if the initial deficit is not too large relative to the gross gain of compliance. Second, we assume that the maximally possible shock S is sufficiently large so that s* ≤ S always holds. This assumption requires the initial deficit to be large enough.
9
Period0.
Wenowturntotheanalysisofperiod0,inwhichthegovernmentchoosesr0andthere‐
fore affects the probability of compliance via (9). From the incumbent government’s
viewinperiod0theutilityisuncertainduetotheshocks.Theexpectedpayoffis
∗
∗
∗
(10)
The first lineshows insquarebrackets theutility (periods1and2)undercompliance
andnon‐compliance,respectively,dependingontherealizationoftheshocks.For low
levelsofs,s≤s*,thegovernmentcompliesinperiod1bydeficitreductionleadingtod2
=0(thefirstintegral).Ifsishigherthans*,thegovernmentdoesnotcomply(thesecond
integral).3Rewritingterms,thesecondlinein(10)showsinbracketsthesameexpres‐
sionasbefore,nowasthesumoftheguaranteedutilityundernon‐complianceandthe
expected gross gain from compliance,minus the cost of deficit reduction in period 1
whensissufficientlysmall(s<s*).
Firstperioddeficit reductionr0affects (10)via thecostofeffort inperiod0 (the first
termin(10)),theprobabilityofrealizingthegrossgainofcompliancep,andthecostof
effortinperiod2inthelattercase.Recallthatthethresholdlevels*isafunctionofr0via
(7)and(9).Thederivativeofexpectedutilitywithrespecttor0is
1
∗
1 ∗∗
(11)
Derivative(11)hasthefollowinginterpretation:Anincreaseinr0increasesthemarginal
costofdeficit reduction in the currentperiod.Themarginalbenefit ofdoing so is the
discounted increase in the expected gross gain of compliance (due to the increase in
probabilityofcompliance)adjustedforthecostofreducingthedeficitbyd0‐r0.Recall
3 This assumes implicitly that d1 >0, which holds, if r0<d0.
10
that (qS)‐1 represents the increase in the probability of compliancewhen r0 is raised
marginally.4
Evaluating(11)attwovaluesofr0providesadditionalinsight:Ontheonehand,atr0=0
themarginalcostofdeficitreductioninperiod0iszerobyassumption,andhencethe
expectedutilitygaindE[U]/dr0ispositivewhenthegrossgainuislargerthanthecostof
reducingtheoriginaldeficitd0(u>c(d0)).Wemakethatassumption,whichinsuresthat
r0≤0cannotbeasolutionto(11)(whensetequaltozero).Atr0=d0,ontheotherhand,
thegovernmentfaceshighmarginalcostinitially,butgainsbyincreasingtheprobability
of compliance (1/(qS))weightedby thegrossgainu.Weassumec’(d0)> δu/(qS), so
thatdE[U]/dr0<0atr0=d0.Thusalocalmaximummustobtainintheintervalbetween
0andd0.Theoptimal levelof firstperioddeficit reduction is foundby setting (11)
equaltozero,whichgives
. (12)
Givenourassumptionsjustmade,thesecondorderconditionisfulfilledat :
0. (13)
2.3Results
We now study the effects of exogenous variables on the probability of compliance p,
whichdependsonexogenousmodelparametersbothdirectly,asshownin(8),butalso
indirectlyviatheoptimallevelofinitialdeficitreductionr0,asimplicitlydefinedin(12).
Thelatteristheperiod0anticipationeffect,whereastheformeristheperiod1compli‐
anceincentiveeffect.
1. Initial deficit: Differentiation of (12) shows that an increase in the initial deficit d0
leadstoalowergovernmenteffortininitialdeficitreduction,thatis
0, (14)
4 The difference between the optimal deficit reduction under credible and non-credible fiscal rules is twofold. First, the utility gain from compliance u appears in (11) but not in (5) because with a credible rule the govern-ment always obtains uc. Second, the marginal benefit of extending r0 does not contain the cost term c(d1) in (11) because an increase in r0 reduces the cost of deficit reduction for given probability of compliance (p or s* in (10)), but at the same time makes compliance more likely (s* goes up). The two effects cancel out.
11
whichisnegativebythesecondordercondition.Theprobabilityofcompliancepisalso
loweredbythedirect(complianceincentive)effectsothatthetotaleffectbecomes
1 0. (15)
Stateswithalargerinitialdeficitarelesslikelytocomplywiththebalancedbudgetre‐
quirementinperiod2(Hypothesis1:H1).
2.Bailoutexpectations: Up to nowwedidnot explicitly address the role of a possible
bailoutincaseofnon‐compliancewiththefiscalrule.Ratherweassignedautilitylevel
forthecaseofnon‐compliance,assumingittobelowerthanincaseofcompliance.Sup‐
posenow that a bailout is possible but less than certain (sonon‐compliance isworse
than compliance in expected terms: unc< uc), and consider that the probability of a
bailoutgoesup.Thisaffectsthegovernmenteffortinreachingthedeficittarget.Formal‐
ly,wecapturethebailoutprobabilitybyinterpretingtheutilityfromnon‐complianceunc
asexpectedutility,whichcomprisestheutilitywhennobailoutoccursandwhenitdoes
occur.Anincreaseinthebailoutprobabilitythusleadstoahigherlevelofunc,andthus
lowernetutilitygainu.Thecomparativestaticsare
0, (16)
thusloweringtheeffortininitialdeficitreduction.Moreover,ahigherbailoututilityre‐
duces theprobabilityof compliancebecauseadecrease inudecreasespbothdirectly
andindirectly:
0. (17)
Weconcludethathigherbailoutexpectationsmakecompliancewiththebalancedbudg‐
etrequirementlesslikely(Hypothesis2:H2).
3.State fiscalrule inperiod1:The fiscalruleunderconsiderationbecomeseffective in
period2.SomestatesinGermanyhaveintroducedfiscalrulesatthestatelevelwithcon‐
straintsbecomingeffectiveprior to thenationaldebtbrake’scrucialyear2020.These
state rules are supposed to strengthen the effort and likelihood of compliance. In the
present frameworkwecapture this ideabyallowing foranadditional fiscalrule tobe
alreadyeffectiveinperiod1.Weassumethattheadditionalfiscalruleiscredible,per‐
12
hapsbecausethereisnoonetobailoutthegovernmentwithinthestate.Yetweallow
for the possibility that the fiscal rule may be of different strictness. We express the
strictness in termsof themaximum feasibledeficit that canoccur inperiod1,d0+qS.
Theupperlimitofthedeficitinperiod1mustobey
. (18)
Theparameterαfrom[0,1]representsthestrengthofthefiscalrule.Thefiscalrulehas
nobitewhatsoeverwhenα=1becausenodeficitreductionisnecessaryinperiod0tobe
compliantwiththenewruleinperiod1.Bycontrast,α=0meansthatthegovernmentis
notallowedtorunagovernmentdeficitinperiod1regardlessofswhenthenewfiscal
ruleiscredible.Thiswouldmandatedeficitreductioninperiod0of ,thusinduc‐
ingd1 0.Lowervaluesofαthuscorrespondtoatighterfiscalruleinperiod1.Using(2)
wecanreformulatetherequirementin(18)intermsofinitialdeficitreduction:
1 : . (19)
Notethat isdecreasinginα.Atighterfiscalruleinperiod1requiresahigherdeficit
reductioneffortinperiod0.Whethertheadditionalfiscalrulehasbitedependsonthe
magnitudesof and ,wherethelatteristakenfrom(12)andrepresentstheoptimal
choiceofinitialdeficitreductionintheabsenceofthefiscalruleinperiod1.When
,thenewfiscalruleisbinding,otherwiseitisnot.Thisresulthasfurtherramifications
fortheprobabilityofcompliancewiththeoriginal fiscalrule inperiod2.Probabilityp
dependspositivelyonr0.
We conclude that the likelihoodof compliance (weakly) increases in the strengthof a
crediblefiscalruleatstatelevelwhichrestrictstheperiod1deficit(Hypothesis3:H3).
4.IndividualBeliefs:Considernowthebeliefsingovernmentcomplianceafterthedeci‐
sion onperiod0 deficit reductionhas been takenbut before the shock s realizes.We
thus focus on the expectations at an interim stage for a given level of r0.Wewish to
compare the beliefs in compliance of two types of politicians: the incumbent govern‐
mentorin‐statelegislaturesontheonehand,andoppositionpoliticiansorout‐of‐state
politiciansontheotherhand.
Thepsychologicalliterature(seeMooreandHealy,2008)suggeststhatalargenumber
of individuals (more than half) believe to perform better than the average/median,
13
which is impossible. This is termed overconfidence. In the present context this could
mean that the incumbentgovernmentbelieves its competency tobehigher thanwhat
theoppositionasserts,thatis,thegovernmentbelievestohavealowervalueofqthan
what theopposition thinks thisvalue tobe.Thisassumptiondoesnotrequireastate‐
mentabout the truecompetence,only that the twobeliefsdiffer.Equation(8) implies
immediately that for given r0 the incumbent’s subjective probability of compliance is
higherthanthatoftheopposition.Thiseffectisreinforcedwhenperiod0deficitreduc‐
tionisendogenous.Toseethis,notethattheeffectfromhighervaluesofqondeficitre‐
ductioninperiod0isfoundbydifferentiationof(12),assumingthatqisthetruevalue:
0, (20)
which is negative by the second order condition. The incumbent government ismore
optimisticaboutthelikelihoodofcomplianceinperiod1andthusundertakesmoreef‐
fortinperiod0,whichinturnmakescompliancemorelikely.
Alternatively,wemayassumethat insiders(whichmaybe the incumbentgovernment
or in‐state legislators) know the government’s competence exactly, denoted by q, but
outsiders (whichmay be the opposition or out‐of‐state politicians) have only a noisy
signal about the government’s competence. Specifically,we assume that outsiders be‐
lieve that government competence is qlwithprobability z andqhwithprobability1‐z
suchthatE[q]=zql+(1‐z)qh=q.Theexpectedvalueoftheoutsiders’subjectivebeliefof
government competence equals therefore the true government competence.We now
comparetheexpectedcomplianceofthegovernmentbycomparingthebeliefsofinsid‐
ersandoutsiders,againataninterimstagewhenr0hasbeensetalready.Theinsider’s
beliefissimply
. (21)
Bycontrast,theoutsider’sexpectedlikelihoodofcompliancebythegovernmentis
1 (22)
.
Comparisonof(21)and(22)showsthatpout>pins.Inotherwords,theoutsiderbelieves
undernoisyinformationthatthegovernmentismorelikelytocomplythantheinsider.
14
Theintuitioncomesfromtheobservationthattheprobabilitypin(8)isaconvexfunc‐
tionofq.
Weconcludethattheoutsiders(politicaloppositionorout‐of‐statepoliticians)aremore
optimisticaboutcompliancethaninsiders(theincumbentorin‐statepoliticians)under
noisy information about the incumbent government’s competence but less optimistic
underoverconfidence(Hypothesis4:H4).
Thus,ourmodelarrivesathypothesesontheheterogeneityofcomplianceexpectations
acrossindividualpoliticians.Thesehypothesesarederivedforasettingwherejurisdic‐
tionsareconfrontedwithanidenticalfiscalrule,asitisthecaseforGermanstatesand
thenationaldebtbrake.OursurveyamongmembersofGermanstateparliamentsoffers
thebasisfortestingtheirrelevance.
3.Institutionalandsurveydetails
3.1.Germany’sfederalsystemandtheconstitutionaldebtbrake
BeforeweintroducethesurveyweprovideabriefintroductiontoGermany’selectoral,
politicalandfiscalsystem(foramoredetaileddescriptionoftheGermanpartyandelec‐
toralsystemthereaderisreferredtoRoberts,1988,andPoguntke,1994).
Democracy. Germany is a parliamentarydemocracywith two chambers at the federal
level:thelowerchambercalledBundestag,whichiselectedbyallcitizens,andtheupper
chambercalledBundesrat,whichrepresentsthe16statesofGermanyandwhosemem‐
bersaredelegatesofstategovernments.Thedebtbrakewasapprovedin2009bymore
thanthe2/3requiredmajorityinbothchambersinordertochangetheconstitution.At
thestatelevel,thereexistsonlyonechamberlikethelowerchamberatthefederallevel.
Wesurveyedmembersofthesestateparliaments,calledMSPhenceforth.
Parties. The number of political parties has some regional variation.We describe the
mainparties:TheChristianDemocrats(CDU/CSU)areacentre‐rightparty,whichpur‐
suesarelativelymarketorientedpolicybutwhichissociallyconservativeinsomestates
(likeBavaria)andonsomepolicyissues(suchasthetraditionalroleofthefamily).The
SocialDemocratic Party (SPD) is the othermajor party and represents the center‐left
(lessmarketorientedthantheChristianDemocrats,sociallyprogressiveandinfavorof
more intense redistribution than CDU/CSU). The Free Democratic Party (FDP) is the
mostmarket oriented partywhich favors small government and low taxes. On social
15
issuesitismoreprogressivethantheChristianDemocrats.TheLeftPartyunitesformer
communists in East Germany (mostly pragmatic) and disappointed Social Democrats
fromtheleftwinginWestGermany(moreideological).TheGreenPartyisalsoonthe
centre‐leftandpushesforenvironmentalandsocialreformswithdiverseviewsoneco‐
nomicissues.Thepartyispopularwithwell‐educatedindividualsfromthemiddleclass.
FiscalFederalism.TheGermanstatefeaturesthreegovernmentlayerswithpartlyover‐
lapping areas of policy responsibility: (1) the federal level, (2) the states, and (3) the
municipallevel.Taxautonomyatthestatelevelisrelativelylow.Revenuesareequalized
toasignificantdegreeacrossstatesandinadditionthroughverticaltaxsharing.Differ‐
ences in state revenuespercapitaare reducedviaa fiscalequalizationsystem,whose
legalfoundationissetinArticle106oftheGermanconstitution(Grundgesetz),accord‐
ing to which material living conditions should be comparable across German states.
ThroughthelargedegreeofrevenuesharingtheGermanfederalsystemisclosertobe‐
ing an example of cooperative fiscal federalism rather than competitive federalism
(Braun, 2007; for details on equalization and tax sharing see also Heinemann et al.,
2013).
FiscalRules.ThefiscalruleistheGermandebtbrake(“Schuldenbremse”),whichbecame
part of theGerman constitution (the “Grundgesetz”) in 2009. Itwasmotivated by the
continuingbuildupofpublicdebtacrossall levelsofgovernmentsincethe1970s.The
new constitutional rule requires the federal government to run a (cyclically adjusted)
budgetdeficitofnomorethan0.35%ofGDPstartingin2016(seeBundesministerium
für Finanzen, 2009 for a detailed description). For German states (“Länder”) the new
ruleismorestringentandrequiresthemtorunazerodeficit(cyclicallyadjusted).The
zerodeficitconstraintforthestatesdoesnotbecomelegallyeffectiveuntilthebudget‐
aryyear2020.Theruleforthefederalgovernmentisaccompaniedwithaspecificplan
detailinghowthestructuraldeficitshallbereducedbetween2011and2015sothatthe
targetisreachedin2016.Forthestates,nospecificpathexistsingeneral.However,five
states (Berlin,Bremen, Saarland, Saxony‐AnhaltundSchleswig‐Holstein) receive “con‐
solidationaids”intotalof€800millionannuallyuntil2019.Inreturntheyarerequired
toreducetheir2010budgetdeficit inequalstepsuntil2020.Asareactiontothenew
national constitutional rule, several states have adjusted their state constitutions or
statebudgetarylawswithrulesechoingorevensharpeninigthenationalrule(forasur‐
veyseeCiagliaandHeinemann,2013).
16
Enforcement.TheStabilityCouncil(“Stabilitätsrat”)hasthetasktosupervisefiscalper‐
formanceandcompliancebothatthefederalandthestatelevel.Itrepresentsthefederal
ministers for financeandeconomicsaswellasall state financeministers.TheCouncil
hasrelativelylittlepowertoenforcefiscalrulesandimprovefiscalperformancebecause
itisnotallowedtoimposemonetarysanctionsdirectly.Inthecaseofthefivestatesre‐
ceiving consolidation aids the Council is entitled to withhold aids in case of non‐
compliance.Non‐monetarysanctionsforallstatesoriginatefromthepossiblepublicity
oftheStabilityCouncil’sstatementsorfrompoliticalcostsmaterializingifastatebudget
isruledasunconstitutionalbytheFederalConstitutionalCourt.
Economic Performance. The lack of comprehensive monetary sanctions and the long
transitoryperiodraiseseriousquestionsaboutthenewrule’scredibility.Inaddition,the
highly diverse fiscal situation of states feeds diverging expectations. Table 1 provides
informationonkeyindicatorsandshowsthelargedifferenceineconomicactivity.GDP
percapitainHamburg,forexample,ismorethantwiceaslargeasinmosteasternstates.
Debt to state GDP is particularly high for the city states of Berlin and Bremen (both
above60%).Oftenhighdebt levelsgohand inhandwith largeprojected fiscaladjust‐
ments,asidentifiedbytheGermanCouncilofEconomicAdvisors’calculationofconsoli‐
dation need (an index ranging from ‐0.6 +3.5,whereBerlin andBremen are near the
maximum).Inthelightofthesefiscalperformancesitissomewhatsurprisingthatcredit
ratings are fairly positive in all states (all in theA range).One explanation consistent
withtheseobservationsisthatbailoutexpectationsexist.Becausetheserankingsappar‐
entlydonotreflectthestrengthofthedebtruleatthestatelevelingreatdetail,thelast
columnofTable1providesanindexforthestringencyofGermanindividualstates’fis‐
calrulesasdevelopedbyCiagliaandHeinemann(2013).Thisindextakesaccountofthe
rule’scontentsandprecision,legalbasisandenforcement.
17
Table1:EconomicandFiscalIndicators
Popula‐tion2011(inmil‐lions)
GDPpercapita2011(inthousandsof€)
TotaldebttoGDPratio2011(in%)
NeedforConsoli‐dation2011‐2020(in%ofGDP)
BondRating2012a
Indexofstringen‐cyofstatedebtrule
FederalGovernment
81.84 44.02 49.79e ‐ AAAd,e
Baden‐Württemberg
10.79 34.89 17.16 0.10 AAAd 0.62
Bavaria 12.60 35.44 6.79 ‐0.60 AAAd 0.48Berlin 3.50 28.95 61.64 3.50 Aa1c 0.65Brandenburg 2.50 22.08 35.77 2.10 Aa1c 0.51Bremen 0.66 42.39 73.63 3.40 ‐ 0.64Hamburg 1.80 52.49 26.86 0.30 ‐ 0.47Hesse 6.09 37.51 17.28 1.30 AAd 0.50Mecklenburg‐WestPomerania
1.63 21.40 29.11 1.70 ‐ 0.46
LowerSaxony 7.91 28.35 25.42 1.30 ‐ 0.55NorthRhine‐Westphalia
17.84 31.88 33.22 1.60 AA‐d 0.45
Rhineland‐Palatinate
4.00 28.31 32.49 1.80 AAAb 0.69
Saarland 1.01 30.10 41.83 2.80 ‐ 0.70Saxony 4.14 22.98 9.99 0.60 AAAd 0.76Saxony‐Anhalt 2.31 22.43 39.84 2.50 AA+d 0.77Schleswig‐Holstein
2.84 25.95 38.57 1.30 AAAb 0.77
Thuringia 2.22 21.66 35.04 2.30 AAAb 0.66Notes: a from http://www.welt.de/finanzen/article107267058/Bundeslaender‐profitieren‐von‐Deutschland‐Bonds.htmllastaccesson23July2013;bFitch;cMoody’s;dS&P,ereferringtofederallevelalone,nottoaggregateforGermany.Needforconsolidation is taken fromSachverständigenrat (2011)and isbasedon theaveragebudgetdeficits from2007 to2010.Itindicatestheextentofconsolidationnecessarytocomplywiththedebtbrakeby2020.Forthatpurpose,ittakesaccountforpensionobligationsandthereductionoftransfersfromthefederallevel(SpecialPurposeGrants)whichwillbothcomeintoeffectuntil2020.TheIndexofstringencyofthedebtrule isnormalizedbetween0and1,wherehighervaluesindicateamorestringentdebtrule(CiagliaandHeinemann,2013).
Hence,boththelegalsettingandthefiscaldivergenceleaveamplespaceforhighlyhet‐
erogeneous expectations on state compliance which we study through our survey
amongmembersofparliament.
18
3.2.Thesurveyamongmembersofstateparliaments
Oursurveywassenttoallmembersofthe16Germanstateparliamentsduringaperiod
of14months in2011and2012.Weconducted the survey in threewaves inorder to
makesurethatitdidnotcollidewithelectiontimes(surveyswereconductedapproxi‐
mately atmid‐term of an electoral cycle).We approachedmembers of parliament by
writtenlettersandsubsequentfollow‐upemails.Ifstillunsuccessful,wecontactedthem
byphone.Takenallthreewavestogether639politiciansfinallyparticipatedinthesur‐
veywhichresultedinaresponserateof34%.Responseratesdifferalongstateandpar‐
tyaffiliation.Table2providesanoverview.Possibleconcernsabouttheeffectofdiffer‐
entresponseratesaredealtwithintheeconometricanalysisbelow.
Table 2: Response rates and survey waves Number
ofMSPsNumberofresponses
Responserate
Surveywavea
Lastelectionbeforesurvey
Overall 1861 639 34.34% Baden‐Württemberg 138 77 55.80% 3 3/2011Bavaria 187 75 40.11% 1 9/2008Berlin 149 30 20.13% 3 9/2011Brandenburg 88 19 21.59% 1 9/2009Bremen 83 18 21.69% 3 5/2011Hamburg 124 39 31.45% 2 2/2011Hesse 114 50 43.86% 2 1/2009Mecklenburg‐WestPomerania 71 17 23.94% 3 9/2011LowerSaxony 152 54 35.53% 1 1/2008NorthRhine‐Westphalia 181 51 28.18% 2 5/2010Rhineland‐Palatinate 101 50 49.50% 3 3/2011Saarland 51 20 39.22% 1 8/2009Saxony 133 45 33.83% 2 8/2009Saxony‐Anhalt 106 47 44.79% 2 3/2011Schleswig‐Holstein 95 29 30.53% 1 9/2009Thuringia 88 36 40.91% 1 8/2009Notes:aThefirstwave(1)tookplaceinMarchandApril2011,thesecondwave(2)tookplaceinDecember2011andJanuary2012,andthethirdwave(3)tookplaceinAprilandMay2012
Thesurveywasnon‐anonymousbutpoliticianswereguaranteedconfidentialityforin‐
dividual responses. Thus, we are able to match the survey responses with personal
characteristicssuchaseducation,committeemembership,etc.frompublicsources(per‐
19
sonalorparliamentarywebsites)andwithstatecharacteristicssuchasGDPpercapita,
debt,needforfiscalconsolidation,etc.(seeTableA1intheappendixforallvariables).
Thequestionnaireconsistedofeightquestionscoveringpreferencesforrevenueauton‐
omy and fiscal equalization, spending preferences aswell as questions related to the
debtbrake(forafulldescriptionseeHeinemannetal.,2014).Forourstudy,wefocuson
thefollowingtwoquestions:
Questioncomplianceexpectation:Whichofthe16Germanstateswillcomplywiththeconstitutionaldebtbrakeasof2020withhighprobability?Eachofthe16statescouldbetickedindividuallyoroptions“all”or“none”couldbecho‐
sen.
Inordertoilluminatetheexpectedimpactofthedebtbrakewealsoaskedforthecon‐
sequencesofnon‐compliance:
Questionconsequencesofnon‐compliance:WhatwillhappenifGermanstatesdonotcomplywiththeconstitutionaldebtbrakeasof2020?(multipleanswerspossible)o Constitutionalcourts(onstateandfederallevels)willenforcebudgetconsolidationo Theconstitutionwillbechangedsoastorelaxthedebtbrakeo Transferpaymentstonon‐complyingstatesaregiven,whichhelptolowerthedeficito There will be sanctions against non‐complying states, e.g., lower transfers within the
federalfiscalequalizationschemeo There will be ordinary legal or constitutional interventions in non‐complying states’
budgetautonomyo Mergerofstateso Nothingwillhappeno Other:___________
Figure1indicatesthatthedebtrulecredibilityisimperfectandcomplianceexpectations
differremarkablyacrossstates.WhileBavariaisseenasanalmostcertaincaseofcom‐
pliance (85%believe it is highly probable) the prospects of the city states ofBremen
(3%) and Berlin (4%) are highly pessimistic. These expectations obviously correlate
closelywith current consolidation needs and debt levels (see Table 1). In addition, a
strongasymmetryemergesforinsider/outsiderexpectationsonfinanciallyweakstates
(see Figure 2with the example forMecklenburg‐West Pomerania):WhileMSPs from
otherstatesarehighlyskeptical,alargemajorityofpoliticiansfromeconomicallyweak‐
erstatesexpecttheirstatetorespectthedebtbrakezerodeficitcapbytheyear2020
(seeTableA2intheappendixforfullinformationoncross‐stateexpectationswhichcon‐
20
firms this asymmetry in general). Figure 3 summarizes the results for the non‐
compliance question: For this scenario, a significant number of politicians expect a
strongroleofconstitutionalcourts(bothfromfederalandstatelevel)toenforceconsol‐
idation or sanction. However, a large fraction of politicians expect the government
budget constraint to be soft due to bailout‐transfers or a relaxation of the strict debt
brake.Overall,thesedescriptivefindingspointtothepossiblerelevanceofourmodel’s
predictionontheroleoftheinitialfiscalsituation,bailoutexpectationsortheexpected
asymmetry between insiders and outsiders.We substantiate themodel’s explanatory
powerinthesubsequentregressionanalysis.
Figure 1: Compliance across states
BB=Brandenburg, BE=Berlin, BW=Baden‐Württemberg, BY=Bavaria, HB=Bremen,HE=Hesse, HH=Hamburg, MV=Mecklenburg‐West Pomerania, NI=Lower Saxony, NW=North Rhine‐Westphalia, RP=Rhineland‐Palatinate, SH=Schleswig‐Holstein,SL=Saarland,SN=Saxony,ST=Saxony‐Anhalt,TH=Thuringia
21
Figure2:ComplianceexpectationMecklenburg‐WestPomerania
BB=Brandenburg, BE=Berlin, BW=Baden‐Württemberg, BY=Bavaria, HB=Bremen,HE=Hesse, HH=Hamburg, MV=Mecklenburg‐West Pomerania, NI=Lower Saxony, NW=North Rhine‐Westphalia, RP=Rhineland‐Palatinate, SH=Schleswig‐Holstein,SL=Saarland,SN=Saxony,ST=Saxony‐Anhalt,TH=Thuringia
Figure3:ConsequencesofNon‐Compliance
22
4.Regressionanalyses
Our database is sufficiently rich to test whether the predictions from our theoretical
model on expectation heterogeneity are consistentwith the observable response pat‐
tern.Ourmodelpredictsthatcomplianceexpectationsofpoliticiansshouldberelatedto
theinitialdeficit,ormoregeneral,theinitialeconomicandfiscalconditionsofthestate
inquestion(H1),theindividualpolitician’sbailoutexpectations(H2),theexistenceand
characteristicsof state ruleswhich complement thenationaldebtbrake (H3), and the
individualpolitician’sinsider/outsiderstatus(duetoeitherasymmetricinformationor
overconfidenceonthesideofincumbents,H4).Wecoverthesefourdimensionsasfol‐
lows(forprecisevariableinformationseeTableA1intheappendix):
- Thestatecharacteristicsmergedtothepolitician’sresponsesincludeGDPpercapita
andtheneedforconsolidation(seeTable1).The lattergivesacomprehensivepic‐
tureofthecurrentfiscalandeconomicconditions(H1).Theneedforconsolidationis
taken from theGermanCouncil ofEconomicAdvisors (Sachverständigenrat,2011)
andreflectstheextenttowhichstatesneedtoconsolidatetheirbudgetsuntil2020
whenthedebtbrakecomesintoeffect.
- Forbailout‐expectations(H2)weexploitthesurveyquestionontheexpectedconse‐
quences of non‐compliance (Figure 3). From this question we construct an index
whichcapturestheindividualperceptionofthestrengthofthebudgetconstraint.A
larger indicatorvaluerepresentstheperceptionofastricterbudgetconstraintand
lowerbailout‐expectations.5
- Fortheexistenceandstringencyofastaterule(H3)wemergedatafromCiagliaand
Heinemann (2013)who develop an index for the stringency of German individual
states’fiscalrules,whichtakesaccountoftherule’scontentsandprecision,legalba‐
sisandenforcement.
- The insider‐outsider‐differentiation (H4) has twodimensions: First,we candistin‐
guish between incumbents as insiders and all others,where “incumbents” are de‐
finedasmembersofoneofthegoverningpartiesintherespectivestate.Second,we
5 Indicator construction is as follows: We add one point if a politician expects one of the “tough” reactions to a state non-complying (i.e. “enforcement through constitutional courts”, “sanctions”, “intervention in budget au-tonomy” or “merger of states”) and subtract one point for each of these reactions which is not expected. Analo-gously, we subtract one point for each of the expected “soft” reactions to a state-non complying (i.e. “change of constitution”, “transfers” or “nothing”) and add one point for each of these reaction which is not expected.
23
cancomparetheexpectationsforaspecificstate’scompliancebetweenin‐stateand
out‐of‐statelegislators.Weincludebothdimensionsinourtesting.
Weenrich this theory‐guidedchoiceofvariables throughthe inclusionof further indi‐
vidual and state controls because a growing empirical literature points to the im‐
portanceofthesevariablesforeconomic,monetaryandfiscalperformance(Besleyetal.,
2011, Göhlmann and Vaubel, 2007,Moessinger, 2014).We take account of the politi‐
cian’s gender, age, education (tertiary degree, type of degree, such as in busi‐
ness/economics),roleinparliament(membershipinbudgetcommittee)andexperience
(numberofyearsinparliament).Interalia,thesevariablesproxydifferencesintheindi‐
vidualinformationlevel.Furthermore,weaddpartydummiestoallowfortheimpactof
ideology. Ideology might influence expectations since perceptions of economic con‐
straintscanbebiasedbystrongideologicalpositions(see,forexample,Heinemannand
Janeba,2011,fortheperceptionofglobalizationconstraintontaxpolicy).Amongstate
controlsweincludeadummyforthosestatesreceivingconsolidationaidandtheextent
offiscalequalizationtransfersreceived.Thesevariablescovertransferdependency.Fi‐
nally,weaddadummyforthepoliticalorientationoftheincumbentgovernmentwhich
allows for the possibility that the incumbent’s political orientation has an impact on
complianceexpectationsfortherespectivestate.
4.1Baselineresults
We estimate a probit model with the compliance expectation as dependent variable
(dummyequals1:Politicianexpectsastatetocomplywiththedebtbrakeasof2020;0:
expectastatenottocomply).Sincewehaveexpectationsof639politicianson16states
wecanexploitatotalof10,224observations.Weclusterstandarderrorsforstatepairs.
Column(1) inTable3summarizesourstartingpointwiththefullsetofcontrolvaria‐
bles.Theresults showthat complianceexpectationsare relatedboth to the individual
andownstatecharacteristicsofrespondents.WeincludefixedeffectsforMSPs’statesof
origintoaccountforthepossibilitythatpoliticiansofparticularstatesmaybemoreor
lessoptimisticingeneral(asitissuggestedbythedescriptiveanalysis,seeTableA2in
theappendix).
Allproxiesrelated toour fourhypothesesarehighly significant. Signsare in linewith
thetheoreticalexpectationsfortheH1‐,H2‐andH3‐relatedindicators:Complianceex‐
24
pectationsforstateswithunfavorablestartingpositions(lowerGDPpercapitaorlarger
needforconsolidation)arelessoptimistic.Thebeliefinbailout‐transfersorotherrelax‐
ationsof the fiscalrule(lower index forstrengthofbudgetconstraint) lowerscompli‐
anceexpectations.Astricterstate‐individualfiscalruleiscorrelatedwithamorefavora‐
bleviewforthisparticularstate.Thesizeoftheeffectsissubstantialjudgedonthebasis
of averagemarginal effects:Aonepercentagepoint increaseof a state’s consolidation
need(H1)lowerstheprobabilitythatthisstateisexpectedtobecompliantbyabout10
percentagepoints.Thedifferencebetweenaverysoft (‐7)andveryhard(+7)percep‐
tionof thebudgetconstraint (H2)amounts toan impactof24percentagepoints.And
thedifference between theweakest (0.45) and strongest (0.78) observable state debt
rule(H3)isassociatedwithaprobabilityincreaseof16percentagepointsthatastateis
predictedtocomply.6H4‐relatedproxiesarehighlysignificantforbothinsider‐outsider‐
dimensions: Insiders (members of parties who form a state’s government/in‐state‐
politicians)aremoreoptimisticthanoutsiders(membersofoppositionparties/out‐of‐
state‐MSPs). The size of the effect is much larger for the in‐state vs. out‐of‐state‐
dimension (21 percentage points) than for the government‐opposition‐distinction (4
percentagepoints).Inthelightofourtheory,thepositivesignofinsiderstatuspointsto
the role of overconfidence as driving insider‐outsider‐asymmetry. If outsiders had an
informationdisadvantage they should sometimes over‐ and sometimesunderestimate
thecompetenceofinsidersbutnotnecessarilybesystematicallymorepessimisticthan
insiders. Therefore, the systematically larger optimism of insiders is consistent with
over‐confidenceratherthanwithnoisyinformation.
Theothercontrolvariablesare important tounderstandtheheterogeneityofexpecta‐
tions, as well. The observed education characteristics do not show up significantly.
Membersof thebudgetcommitteeviewadherencetothedebtbrakeasmoredifficult.
Equally,alongerparliamentaryexperiencereducescomplianceexpectation.Thisfinding
isnotdrivenbyanageeffectwhichpointsintotheoppositedirectionwitholdermem‐
bersbeingmoreconfident.Femalelegislatorsaremorepessimisticthantheirmalecol‐
leagues.Partyimprintoncomplianceexpectationsismoderate:Forexample,thereare
nosignificantdifferencesbetweenpartiesfromtheoppositeendsofthepoliticalspec‐
6 (0.78-0.45)*49 = 16 where 0.78 is the largest observed value of the index and 0.45 is the smallest observed value.
25
trum (i.e. between themarket‐liberal FDP and the socialist Left Party).7 Stateswith a
government consisting of right parties (i.e. ChristianDemocrats and/or FDP) are per‐
ceivedtohaveahigherchanceofcompliance.Consolidationaiddoesnotseemtocom‐
pensate for the less favorableeconomicand fiscal conditionsof the five related states
sincetherelateddummyissignificantlynegative.
Weemployvariousmodelvariants:Incolumn(2)weallowforindividualfixedeffects.8
Thisspecificationaccounts for therisk thatunobserved individualcharacteristicsmay
bias the results for state indicators. No substantial differences emerge. Table 4 takes
accountofthespatialdimensionofourcross‐stateanalysis. Inparticular,weallowfor
moredifferentiatedinformationasymmetriesacrossstates.Forthatpurpose,weinclude
twomeasuresof geographicproximitybetweenstates:First thedistancebetween the
ownstate’scapitalandthatofthestatetobeassessedandsecondadummyforacom‐
monborder.Thehypothesisisthatproximityandacommonbordermatterformutual
information.Again,column(1)includesourfullsetofindividualcontrolswhilecolumn
(2)replacesthembyindividualfixedeffects.Onlythedistanceindicatoris(weakly)sig‐
nificantintheindividualfixedeffectsspecifications.Complianceexpectationstendtobe
moreoptimisticformoredistantstatesthanforcloseneighbors.Thisisinlinewiththe
predictionof thetheoreticalmodelwithrespect to informationasymmetries.Allother
resultshardlychange,neitherintermsofsignificancenorinthesizeofmarginaleffects.
4.2Robustnessofregressionresults
First, theresultspresentedabovearerobustwithrespecttotheuseofdifferentvaria‐
blescapturingstatefiscalconditions(Hypothesis1).Nomatterwhetherweincludeei‐
therastate’stotaldebtstockrelativetoitsGDPortheaveragebudgetdeficit(overthe
lastthreeyears)relativetoGDPinsteadoftheneedforconsolidation,ourabovefindings
areconfirmed(seeTableA3 in theAppendix). Just like theneed for consolidation the
debt stock and the average deficit enter highly significantly andwith a negative sign.
Higherdebtordeficitsalsodecreasethecomplianceexpectationsoflegislators.Theim‐
pactofalmostallothervariablesremainsasinthebaselineregressions.Thecoefficient
for GDP per capita becomes significant, thereby providing additional evidence for the
7 Weighted regressions, however, indicate that Left Party politicians are more confident that the debt brake will be respected than politicians from the FDP, see below section 4.2. 8 Due to perfect collinearity of individual and home state fixed effects, we have to exclude the latter in this speci-fication.
26
validityofamoregeneralformulationofhypothesis1:Thehighertheincomeperinhab‐
itant,thelessfinanciallyconstrainedisastateandthehighertheprobabilityofcompli‐
ance. Only the coefficients to the fiscal equalization transfers change significance and
signsacrossspecifications.Webelief thatthiscanbeexplainedbythefactthatdebt is
highly correlatedwith financial equalization transfers9,whereas the average deficit is
not.
Second,aconcernaboutthevalidityofourdatacouldoriginatefromsampleselection.
Foroursurvey,Heinemannetal. (2013)haveconductedaunitnon‐responseanalysis.
Theymakeuseofdataonthepersonalcharacteristicsforall1683legislators,notonly
thosewhoresponded.10Thenon‐responseanalysisidentifiesvariablesattheindividual
andstatelevelthataffectpoliticians’participationdecision.Accordingtotheseresults,
significantdriversofsurveyparticipationare:education(degreeineconomicsorbusi‐
ness),budgetcommitteemembership,membershipingovernmentcoalitionpartiesand
gender.Thus, our regressions comprise as controls those factorswhichare important
drivers of non‐response. This greatly reduces the potential for selection bias. Yet,we
cannot fully exclude that we might still have a selection bias (Little and Vartivarian,
2005).Asa furtherrobustnesscheck,wethereforeemployaweightedregression(see
TableA4 in theappendix).For theweighting,weuse the inverseresponseprobability
basedonpartyandstateaffiliation.Theweightedregressionslightlychangesthefind‐
ings for party dummies: The Social Democrats dummy loses significancewhereas the
differencebetweentheLeftPartyandtheFreeDemocratsnowbecomessignificant.In‐
terestingly, in this regressionvariant left‐ leaningpoliticiansaremoreoptimistic than
their right‐leaning colleagues. The essential findings for our key hypotheses are con‐
firmed.Comparedtothenon‐weightedregressionthereareonlyminorchanges inthe
sizeofaveragemarginaleffects.
9 The correlation coefficient amounts to 0.76. 10 We do not face severe item non-response but predominantly unit non-response. Item non-response amounts to less than 1% of respondents and is therefore negligible for the survey at hand.
27
Table3: LikelihoodofState’sCompliance–ResultsExcludingGeographicProximity
Probitregressionswithcomplianceexpectation asdependentvariable (1:complianceexpected,0:notexpected) (1) (2)
IndependentVariables Baseline1
Averagemarginaleffects Baseline2
Averagemarginaleffects
Individidual:education Tertiarydegree 0.020 0.006
[0.035] [0.009]
Economics/Businessdegree 0.040 0.011 [0.039] [0.011]
Individual:parliamentaryrole Memberofgoverningpartiesinstate(H4) 0.149*** 0.040***
[0.046] [0.013]
Memberofbudgetcommittee ‐0.150*** ‐0.041*** [0.039] [0.011]
Numberofyearsinparliament ‐0.006** ‐0.002** [0.002] [0.001]
Individual:other Female ‐0.106*** ‐0.029***
[0.032] [0.009]
Ageinyears 0.002* 0.001* [0.001] [0.000]
Individual:bailout‐expectation Indexforperceivedstrengthofbudgetconstraint(H2) 0.063*** 0.017***
[0.005] [0.001]
Individual:partyaffiliationa CDU/CSU ‐0.112 ‐0.030
[0.068] [0.019]
SPD ‐0.181** ‐0.049** [0.074] [0.020]
GreenParty 0.050 0.013 [0.087] [0.024]
LeftParty 0.109 0.030 [0.084] [0.023]
OtherParties ‐0.109 ‐0.030 [0.125] [0.034]
Statecharacteristicsb GDPpercapita(H1) 0.007 0.002 0.005 0.001
[0.005] [0.001] [0.007] [0.001]
Needforconsolidation(H1) ‐0.386*** ‐0.105*** ‐0.541*** ‐0.098*** [0.048] [0.013] [0.063] [0.011]
Debtruleindex(H3) 1.860*** 0.506*** 2.736*** 0.497*** [0.292] [0.079] [0.398] [0.072]
Dummyforconsolidationassistance ‐0.767*** ‐0.209*** ‐1.220*** ‐0.222***[0.110] [0.030] [0.147] [0.027]
FiscalequalizationtransferstoGDP ‐0.067 ‐0.018 ‐0.151** ‐0.027** [0.047] [0.013] [0.070] [0.013]
Governmentcoalitionconsistsofrightparties 0.605*** 0.165*** 0.870*** 0.158***[0.074] [0.020] [0.103] [0.018]
Crossstatedimension: Ownstate(H4) 0.759*** 0.207*** 1.121*** 0.204*** [0.104] [0.028] [0.169] [0.030]
Distance ‐‐ ‐‐ ‐‐ ‐‐
Adjacency ‐‐ ‐‐ ‐‐ ‐‐
Homestatefixedeffects Personfixedeffects Regressiondiagnostics: Observations 10,224 10,224Pseudo‐R2 0.243 0.491p‐valuejointsignificanceofallvariables 0.000 0.000p‐valuejointsignificanceofallindividualvariables 0.000 n.a.p‐valuejointsignificanceofparty‐dummies 0.000 n.a.p‐valuejointsignificanceofstatecontrols 0.000 0.000p‐valuejointsignificanceofcrossstatevariables n.a. n.a.Notes:*/**/***denotesignificanceatthe10%/5%/1%level;Standarderrorsinbrackets;abasecategoryisthemarketorientedliberaldemocraticparty“FDP”;bStatecharacteristicsare2010dataforsurveywaves1and2,whichbothtookplacein2011,and2011dataforsurveywave3,whichtookplacein2012.
28
Table4: LikelihoodofState’sCompliance–ResultsIncludingGeographicProximity
Probitregressionswithcomplianceexpectationas dependentvariable(1:complianceexpected,0:notexpected) (1) (2)
IndependentVariables Baseline1
Averagemarginaleffects Baseline2
Averagemarginaleffects
Individidual:education Tertiarydegree 0.021 0.006
[0.035] [0.009]
Economics/Businessdegree 0.039 0.011 [0.039] [0.011]
Individual:parliamentaryrole Memberofgoverningpartiesinstate(H4) 0.148*** 0.040***
[0.046] [0.012]
Memberofbudgetcommittee ‐0.150*** ‐0.041*** [0.039] [0.011]
Numberofyearsinparliament ‐0.006** ‐0.002** [0.002] [0.001]
Individual:other Female ‐0.105*** ‐0.029***
[0.032] [0.009]
Ageinyears 0.002* 0.001* [0.001] [0.000]
Individual:bailout‐expectation Indexforperceivedstrengthofbudgetconstraint(H2) 0.063*** 0.017***
[0.005] [0.001]
Individual:partyaffiliationa CDU/CSU ‐0.112 ‐0.030
[0.069] [0.019]
SPD ‐0.181** ‐0.049** [0.074] [0.020]
GreenParty 0.050 0.014 [0.087] [0.024]
LeftParty 0.108 0.029 [0.084] [0.023]
OtherParties ‐0.107 ‐0.029 [0.126] [0.034]
Statecharacteristicsb GDPpercapita(H1) 0.007 0.002 0.005 0.001
[0.005] [0.001] [0.007] [0.001]
Needforconsolidation(H1) ‐0.408*** ‐0.111*** ‐0.572*** ‐0.103*** [0.048] [0.013] [0.064] [0.011]
Debtruleindex(H3) 1.806*** 0.491*** 2.683*** 0.485*** [0.286] [0.078] [0.393] [0.071]
Dummyforconsolidationassistance ‐0.723*** ‐0.196*** ‐1.166*** ‐0.211***[0.112] [0.030] [0.150] [0.028]
FiscalequalizationtransferstoGDP ‐0.045 ‐0.012 ‐0.116* ‐0.021* [0.046] [0.012] [0.069] [0.012]
Governmentcoalitionconsistsofrightparties 0.582*** 0.158*** 0.838*** 0.151***[0.071] [0.019] [0.099] [0.017]
Crossstatedimension Ownstate(H4) 0.621*** 0.169*** 0.940*** 0.170*** [0.129] [0.035] [0.197] [0.036]
Distance ‐0.014 ‐0.004 ‐0.010 ‐0.002 [0.026] [0.007] [0.037] [0.007]
Adjacency 0.133 0.036 0.217* 0.039* [0.083] [0.022] [0.119] [0.021]
Homestatefixedeffects Personfixedeffects Regressiondiagnostics: Observations 10,224 10,224Pseudo‐R2 0.244 0.496p‐valuejointsignificanceofallvariables 0.000 0.000p‐valuejointsignificanceofallindividualvariables 0.000 n.a.p‐valuejointsignificanceofparty‐dummies 0.000 n.a.p‐valuejointsignificanceofstatecontrols 0.000 0.000p‐valuejointsignificanceofcrossstatevariables 0.000 0.000Notes:*/**/***denotesignificanceatthe10%/5%/1%level;Standarderrorsinbrackets;abasecategoryisthemarketorientedliberaldemocraticparty“FDP”;bStatecharacteristicsare2010dataforsurveywaves1and2,whichbothtookplacein2011,and2011dataforsurveywave3,whichtookplacein2012.
29
5.Conclusion
Fiscalrulesaredesignedtoinfluencefiscalperformanceofstates.Whenafiscalruleis
effective,itmustimpactontheexpectationsandbeliefsofthosepoliticianswhodecide
onthegovernmentbudget.OurstudyofthenewdebtbrakeinGermanyrevealsanim‐
perfectcredibilityofthefiscalruleandpointstohighlyheterogeneousexpectationswith
respecttosub‐nationalcompliance.
Anessentialresultrelatestotheasymmetricexpectationsofinsidersandoutsiders.This
holds both for the government versus opposition and the in‐state versus out‐of‐state
dimension.Thisresultmightbeconsideredunproblematic,ifthegoverningpartiesand
politicians in the state under considerationwere better informed and thereforemore
trustworthyintheirjudgmentsthanoutsiders.Ourempiricalfindingsbasedonatheo‐
retical model point into a different direction, however. Insiders (in state politicians,
membersfromgoverningcoalitionparties)aremoreoptimisticthanoutsidersandare
likelytobesubjecttoanoverconfidencebias,whichcouldleadtotoolittleconsolidation
effort.Theasymmetryhas thepotential tounderminea fiscal rule’seffectiveness:The
prevalentexpectationthatother jurisdictionsmightnotcomplycouldalsoweakenthe
perceivedpressurefortheownstate.
Ouranalysisallowsustodrawafewtentativeconclusionsthatshouldbetakenintoac‐
countinthedesignoffiscalrulesalsointheEuropeancontext.First,aweakinitialfiscal
situationisaburdenforrulecredibility.Thephasing‐inofanewruleshouldbeparal‐
leledbyattemptstoremoveoratleastreducetheproblemofunsustainablebudgetary
legaciessuchashighinitialdebt.Second,sub‐nationalrulesmightbeahelpfulcomple‐
menttoanationalruleinafederalcontextlikeGermanywherestateshavesubstantial
spendinganddeficitautonomy.Andthird,clearandcomprehensivesanctionsandcon‐
sequencesincaseofnon‐complianceareimportanttoanchorcomplianceexpectations.
30
References
Azzimonti,M.,Battaglini,M.andS.Coate "AnalyzingtheCaseforaBalancedBudgetAmendmenttotheU.S.Constitution",mimeo,2008.
Barro,R. J.andD.B.Gordon"APositiveTheoryofMonetaryPolicy inaNatural‐RateModel",JournalofPoliticalEconomy91(4),1983,589–610.
Besley,T.,Montalvo,J.G.andM.Reynal‐Querol“DoEducatedLeadersMatter?”,TheEconomicJournal121,2011,205–227.
Braun,D.“HowtoMakeGermanFiscalFederalismSelf‐Enforcing:AComparativeAna‐lysis“,ZeitschriftfürStaats‐undEuropawissenschaften2,2007,235‐262.
Buchanan, J.M. "Lagged implementationasanelement inconstitutional strategy",Eu‐ropeanJournalofPoliticalEconomy10,1994,11‐26.
Bundesministerium der Finanzen “Reform der verfassungsrechtlichen Verschul‐dungsregeln von Bund und Ländern, Die Empfehlungen der Föderalismusreform II“,MonthlyReportMarch2009,36‐44.
Ciaglia,S.andF.Heinemann“DebtRuleFederalism:TheCaseofGermany”,ZeitschriftfürStaats‐undEuropawissenschaften11(4),2013,570‐602.
Dahan,M. andM. Strawczynski “Fiscal Rules and Composition Bias in OECD Coun‐tries”,CESifoWorkingPaperSeries3088,Munich,2010.
Debrun,X. “Fiscal rules in amonetary union: a short‐run analysis”, Open EconomiesReview11(4),2010,323–358.
Debrun,X.,Moulin, L.,Turrini,A.,Ayuso‐i‐Casals, J. andM.S.Kumar “Tied to themast?NationalfiscalrulesintheEuropeanUnion”,EconomicPolicy23(54),2008,297–362.
Eichengreen,B.andT.Bayoumi“Thepoliticaleconomyoffiscalrestrictions:implica‐tions forEurope from theUnited States”, EuropeanEconomicReview38 (3–4), 1994,783–791.
EuropeanCouncil"StatementbytheEuroAreaHeadsofStateorGovernment,Brussels,December9,2011.
Feld,L.P.andG.Kirchgässner"OntheEffectivenessofDebtBrakes:TheSwissExperi‐ence", inR.Neckand J.‐E. Sturm (eds.), SustainabilityofPublicDebt, Cambridge,USA,MITPress,2008,223‐255.
Fink,A. andT.Stratmann“InstitutionalizedBailoutsandFiscalPolicy:ConsequencesofSoftBudgetConstraints”,mimeoApril2010.
Göhlmann,S.andR.Vaubel“TheEducationalandOccupationalBackgroundofCentralBankersandItsEffectonInflation:AnEmpiricalAnalysis”,EuropeanEconomicReview51(4),2008,925‐941.
Goodspeed,T. J. “Bailouts in a Federation”, International Tax and Public Finance, 9,2002,409–421.
Heinemann,F.andS.Osterloh “ThePoliticalEconomyofCorporateTaxHarmoniza‐tion‐WhyDoEuropeanPoliticians(Dis)likeMinimumTaxRates?”EuropeanJournalofPoliticalEconomy29,2013,18‐37.
31
Heinemann,F.,Osterloh, S.andA.Kalb “Sovereign Risk Premia: The Link betweenFiscalRulesandStabilityCulture”,JournalofInternationalMoneyandFinance41,2014,110‐127.
Heinemann, F. and E. Janeba “Viewing Tax Policy Through Party‐Coloured Glasses:WhatGermanPoliticiansBelieve”,GermanEconomicReview12(3),August2011,286‐311.
Heinemann,F., Janeba,E.,Moessinger,M.‐D.,Schröder,C.andF.Streif “Föderalis‐mus‐PräferenzenindendeutschenLandesparlamenten“,PerspektivenderWirtschafts‐politik,15(1),2014,56‐74.
Heinemann,F., Janeba,E.,Moessinger,M.‐D.andC.Schröder “RevenueAutonomyPreferenceinGermanStateParliaments”,ZEWDiscussionPaper,2013,No.03‐090.
Iara,A.andG.Wolff “RulesandRisk in theEuroArea”,European JournalofPoliticalEconomy34,2014,222‐236.
InternationalMonetary Fund “Balancing Fiscal Policy Risks”, Fiscal Monitor, April2012.
Janeba,E.andS.Osterloh“TaxandtheCity–Atheoryoflocaltaxcompetitionandevi‐denceforGermany”,JournalofPublicEconomics106,2013,89‐100.
Janeba, E. “Germany’s New Debt Break Rule: A Blueprint for Europe?” Finanzar‐chiv/PublicFinanceAnalysis68(4),2012,383‐405.
Kirchgässner,G. ”TheEffectsofFiscal InstitutionsonPublicFinance:ASurveyof theEmpiricalEvidence”, in:S.L.WinerandH.Shibata (eds.),PoliticalEconomyandPublicFinance:TheRoleofPoliticalEconomyintheTheoryandPracticeofPublicEconomics,EdgarElgar,Cheltenham2002,S.145‐177.
Krogstrup,S.andS.Wälti “Do fiscal rulescausebudgetaryoutcomes?”PublicChoice136(1),2008,123–138.
Kydland, F.E. and E.E.Prescott “Rules Rather than Discretion: The Inconsistency ofOptimalPlans”,TheJournalofPoliticalEconomy,85/3,1977,473‐492.
Lagona,F.andF.Padovano“Anonlinearprincipalcomponentanalysisoftherelation‐shipbetweenbudgetrulesandfiscalperformanceintheEuropeanUnion,”PublicChoice130(3–4),2007,401–436.
Little,RandS.Vartivarian“DoesWeightingforNonresponseIncreasetheVarianceofSurveyMeans?”,Michigan,September2005.
Moessinger,M.‐D.“DothePersonalCharacteristicsofFinanceMinistersAffectChangesofPublicDebt?”,PublicChoice,2014,DOI10.1007/s11127‐013‐0147‐x.
Moore,D.A.and J.P.Healy “TheTroublewithOverconfidence”,PsychologicalReview115,2008,502‐517.
Poguntke,T.”PartiesinaLegalisticCulture:TheCaseofGermany”,in:R.S.KatzandP.Mair(eds.),HowPartiesOrganize,SagePublications,London,1994,185–215.
Poterba,J.M.“BudgetinstitutionsandfiscalpolicyintheU.S.states”,AmericanEconom‐icReview86(2),1996,395–400.
Roberts,G. ‘TheGerman Federal Republic: TheTwo‐LaneRoute toBonn’, in:M. Gal‐lagherandM.Marsh(eds.),CandidateSelectioninComparativePerspective,Sage
32
Publications,London,1988,94–118.
Sachverständigenrat “Verantwortung für Europa wahrnehmen – Jahresgutachten2011/12“,SachverständigenratzurBegutachtungdergesamtwirtschaftlichenEntwick‐lung,Wiesbaden,2011.
Voigt, S. and L.Blume “The Economic Effects of Constitutional Budget Institutions”,mimeo.2011.
33
AppendixA.TablesTableA1:IndividualandStateVariables
Variable Unit Explanations Individual:education
Tertiarydegree Dummy Degreefromuniversityorpolytechnic
Economics/Businessdegree Dummy Tertiaryeducationinbusinessoreconomics
Individual:parliamentaryrole
Memberofgoverningpar‐tiesinstate
Dummy Memberofoneoftherulingparties
Memberofbudgetcommit‐tee
Dummy Dealswithstategovernmentbudget
Numberofyearsinparlia‐ment
Discrete Calculatedas2011/2012minusyearofparliamententry(interruptionstakenintoaccount)
Individual:other
Female Dummy Memberofparliamentisfemale
Ageinyears Discrete Calculatedas2011/2012minusyearofbirth
Individual:bailout‐expectation
Indexforperceivedstrengthofbudgetconstraint
Discrete Measurerangingfrom‐7to+7,withhighervaluesindicatingahigherexpectationofthedebtbrakebe‐ingenforcedincaseofnon‐compliance,seefootnote3
Individual:partyaffiliation
CDU/CSU Dummy MemberofChristianDemocraticorChristianSocialParty
FDP Dummy MemberofFreeDemocraticParty
GreenParty Dummy MemberofGreenParty
LeftParty Dummy MemberofLeftParty(notincludedintoregressionssinceitservesasbasecategory)
SPD Dummy MemberofSocialDemocraticParty
Other Dummy MemberofotherParty
Statecharacteristics
GDPpercapita Continuous Grossdomesticproductpercapita,inthousandsofEuros,source:GermanStatisticalOffice
Needforconsolidation Continuous In%ofGDP,consolidationneededtocomplywithdebtbrakebytheyear2020,source:Sachverstän‐digenrat(2011)
TotaldebttoGDP Continuous Totaldebtdividedbygrossdomesticproduct,in%,source:GermanStatisticalOffice
ThreeyearaveragebudgetdeficittoGDP
Continuous Weightedaverageofthelastthreebudgetdeficitsdividedbygrossdomesticproduct,in%,source:GermanStatisticalOffice
Indexofstringencyofstatedebtrule
Continuous
Normalizedbetween0and 1,largervaluesindicatingstricterrule,source:CiagliaandHeinemann(2013)
Dummyforconsolidation Dummy Takesthevalueof1forstatesreceivingconsolidation
34
assistance assistance
Fiscalequalizationtransferstototalspending
Continuous Totalnetintra‐statetransferpaymentsdividedbytotalspending,in%,sources:FederalMinistryofFinance,GermanStatisticalOffice
Governmentcoalitioncon‐sistsofrightparties
Dummy Takesthevalueof1forapurelyright‐leaninggov‐ernment(coalition),avalueof0.5foramixedgov‐ernmentcoalitionandavalueof0forapurelyleft‐leaninggovernment(coalition)
Crossstatedimension
Distance Continuous Distancein100kmbetweenanytwostatecapitalcities
Adjacency Dummy Takesonthevalueof1 ifthehomestateofthere‐spondentandthestatetobeevaluatedshareacom‐monborder(andifthestatetobeevaluatedisthehomestateoftherespondent)
Ownstate Dummy Takesonthevalueof1 ifthestatetobeevaluateditthehomestateoftherespondent
TableA2:Cross‐statecomplianceexpectations
Evaluatedstates BB BE BW BY HB HE HH MV NI NW RP SH SL SN ST TH ∅
Evaluatingstates
BB 53 5 68 89 0 58 53 11 37 16 32 5 11 68 16 37 35BE 13 33 70 73 0 67 37 30 47 10 23 13 10 57 27 50 35BW 5 0 75 93 1 58 22 16 17 9 19 8 4 71 5 19 26BY 3 3 57 89 3 53 21 4 25 5 17 7 4 61 5 32 24HB 11 0 67 72 11 56 28 28 50 11 28 6 6 56 28 22 30HE 10 2 56 76 2 78 26 18 34 10 16 8 8 58 16 32 28HH 21 8 72 74 0 62 67 31 44 15 36 8 3 54 21 28 34MV 6 0 72 78 0 53 41 83 24 0 12 0 6 78 12 29 31NI 4 0 74 91 2 57 24 19 56 11 26 11 6 54 20 26 30NW 6 4 67 82 0 53 10 24 45 16 29 10 4 61 20 31 29RP 14 0 76 78 4 64 28 20 36 14 52 12 4 64 22 40 33SH 10 7 65 86 10 55 17 24 38 10 21 66 10 52 28 31 33SL 20 5 95 100 5 85 45 20 55 10 35 20 30 55 25 35 40SN 11 0 67 80 2 42 11 29 20 0 13 4 0 89 16 42 27ST 24 3 76 83 7 52 28 45 35 14 28 17 17 72 59 45 38TH 22 11 67 97 11 69 22 31 47 19 33 28 11 89 28 47 40
∅MSP∅State
12 4 69 85 3 59 27 23 36 10 26 13 7 65 19 33 3115 5 70 84 4 60 30 27 38 11 26 14 8 65 22 34 32
#oftimeswhereout‐sidersaremoreopti‐misticthaninsiders
0 0 3 4 0 1 0 0 0 1 0 0 0 0 0 1
Note:FiguresareinpercentandindicatetheshareofMSPswhoexpectthattheevaluatedstatewillbecompliant.∅MSPindicatestheaverageoverallMSPs.∅Stateindicatestheunweightedaverageoverthestatefigures.
35
TableA3: LikelihoodofAnyState’sCompliance–RobustnessChecks1(alternativevar‐iables)
Probitregressionswithcomplianceexpectationasdependentvariable(1:complianceexpected,0:notexpected) (1) (2)
IndependentVariables
Baseline1(withtotaldebt)
Averagemarginaleffects
Baseline1(withbudgetdeficit)
Averagemarginaleffects
Individidual:education Tertiarydegree 0.023 0.006 0.023 0.006
[0.035] [0.009] [0.035] [0.010]
Economics/Businessdegree 0.038 0.010 0.037 0.010[0.039] [0.011] [0.039] [0.011]
Individual:parliamentaryrole Memberofgoverningpartiesinstate(H4) 0.127*** 0.034*** 0.106** 0.029**
[0.046] [0.012] [0.049] [0.013]
Memberofbudgetcommittee ‐0.152*** ‐0.041*** ‐0.149*** ‐0.041***[0.039] [0.011] [0.039] [0.011]
Numberofyearsinparliament ‐0.006*** ‐0.002*** ‐0.006*** ‐0.002*** [0.002] [0.001] [0.002] [0.001]
Individual:other Female ‐0.105*** ‐0.028*** ‐0.106*** ‐0.029***
[0.033] [0.009] [0.032] [0.009]
Ageinyears 0.002* 0.001* 0.002* 0.001*[0.001] [0.000] [0.001] [0.000]
Individual:bailout‐expectation Indexforperceivedstrengthofbudgetconstraint(H2) 0.063*** 0.017*** 0.062*** 0.017***
[0.005] [0.001] [0.005] [0.001]
Individual:partyaffiliationa CDU/CSU ‐0.110 ‐0.030 ‐0.103 ‐0.028
[0.068] [0.018] [0.067] [0.018]
SPD ‐0.177** ‐0.048** ‐0.167** ‐0.046** [0.074] [0.020] [0.073] [0.020]
GreenParty 0.047 0.013 0.046 0.013 [0.088] [0.024] [0.087] [0.024]
LeftParty 0.103 0.028 0.097 0.026[0.085] [0.023] [0.085] [0.023]
OtherParties ‐0.115 ‐0.031 ‐0.117 ‐0.032[0.128] [0.034] [0.126] [0.034]
Statecharacteristicsb GDPpercapita(H1) 0.022*** 0.006*** 0.027*** 0.007***
[0.005] [0.001] [0.005] [0.001]
TotaldebttoGDP(H1) ‐0.053*** ‐0.014*** [0.005] [0.001]
Averagebudgetdeficitoverlastthreeyears(H1) ‐0.314*** ‐0.086*** [0.034] [0.009]
Debtruleindex(H3) 0.896*** 0.241*** 2.581*** 0.706*** [0.324] [0.087] [0.285] [0.077]
Dummyforconsolidationassistance ‐0.100 ‐0.027 ‐0.734*** ‐0.201***[0.117] [0.031] [0.094] [0.025]
FiscalequalizationtransferstoGDP 0.148*** 0.040*** ‐0.315*** ‐0.086*** [0.053] [0.014] [0.051] [0.014]
Governmentcoalitionconsistsofrightparties 0.145* 0.039* 0.590*** 0.161***[0.075] [0.020] [0.073] [0.020]
Crossstatedimension Ownstate(H4) 0.627*** 0.169*** 0.731*** 0.200*** [0.106] [0.028] [0.129] [0.035]
Distance ‐0.015 ‐0.004 0.011 0.003 [0.021] [0.006] [0.025] [0.007]
Adjacency 0.147** 0.040** 0.136* 0.037* [0.073] [0.020] [0.083] [0.023]
Homestatefixedeffects Regressiondiagnostics: Observations 10,224 10,224Pseudo‐R2 0.253 0.242p‐valuejointsignificanceofallvariables 0.000 0.000p‐valuejointsignificanceofallindividualvariables 0.000 0.000p‐valuejointsignificanceofparty‐dummies 0.000 0.000p‐valuejointsignificanceofstatecontrols 0.000 0.000p‐valuejointsignificanceofcrossstatevariables 0.000 0.000Notes:*/**/***denotesignificanceatthe10%/5%/1%level;Standarderrorsinbrackets;abasecategoryisthemarketorientedliberaldemocraticparty“FDP”;bStatecharacteristicsare2010dataforsurveywaves1and2,whichbothtookplacein2011,and2011dataforsurveywave3,whichtookplacein2012.
36
TableA4: LikelihoodofAnyState’sCompliance–RobustnessChecks2(Weighting)
Probitregressionswithcomplianceexpectationasdependentvariable(1:complianceexpected,0:notexpected) (1) (2)
IndependentVariables
Baseline1(Weightedregression)
Averagemarginaleffects
Baseline1(Weightedregression)
Averagemarginaleffects
Individidual:education Tertiarydegree 0.008 0.002 0.009 0.002
[0.039] [0.011] [0.040] [0.011]
Economics/Businessdegree 0.077* 0.021* 0.076* 0.021*[0.042] [0.012] [0.042] [0.012]
Individual:parliamentaryrole Memberofgoverningpartiesinstate(H4) 0.138*** 0.038*** 0.140*** 0.039***
[0.047] [0.013] [0.047] [0.013]
Memberofbudgetcommittee ‐0.140*** ‐0.039*** ‐0.140*** ‐0.039***[0.041] [0.011] [0.041] [0.011]
Numberofyearsinparliament ‐0.015*** ‐0.004*** ‐0.015*** ‐0.004*** [0.003] [0.001] [0.003] [0.001]
Individual:other Female ‐0.084** ‐0.023** ‐0.084** ‐0.023**
[0.036] [0.010] [0.036] [0.010]
Ageinyears 0.007*** 0.002*** 0.007*** 0.002***[0.002] [0.001] [0.002] [0.001]
Individual:bailout‐expectation Indexforperceivedstrengthofbudgetconstraint(H2) 0.057*** 0.016*** 0.057*** 0.016***
[0.006] [0.002] [0.006] [0.002]
Individual:partyaffiliationa CDU/CSU 0.000 0.000 ‐0.001 ‐0.000
[0.075] [0.021] [0.075] [0.021]
SPD ‐0.125 ‐0.035 ‐0.126 ‐0.035 [0.082] [0.023] [0.082] [0.023]
GreenParty 0.037 0.010 0.037 0.010 [0.097] [0.027] [0.097] [0.027]
LeftParty 0.287*** 0.080*** 0.287*** 0.079***[0.098] [0.027] [0.098] [0.027]
OtherParties ‐0.525** ‐0.146*** ‐0.523** ‐0.145**[0.204] [0.057] [0.204] [0.056]
Statecharacteristicsb GDPpercapita(H1) 0.004 0.001 0.005 0.001
[0.005] [0.001] [0.005] [0.001]
Needforconsolidation(H1) ‐0.384*** ‐0.107*** ‐0.393*** ‐0.109*** [0.045] [0.012] [0.046] [0.013]
Debtruleindex(H3) 1.770*** 0.492*** 1.763*** 0.489*** [0.286] [0.079] [0.284] [0.078]
Dummyforconsolidationassistance ‐0.733*** ‐0.203*** ‐0.713*** ‐0.198***[0.108] [0.029] [0.112] [0.030]
FiscalequalizationtransferstoGDP ‐0.049 ‐0.014 ‐0.034 ‐0.009 [0.045] [0.012] [0.044] [0.012]
Governmentcoalitionconsistsofrightparties 0.578*** 0.160*** 0.558*** 0.155***[0.072] [0.019] [0.070] [0.019]
Crossstatedimension Ownstate(H4) 0.816*** 0.227*** 0.747*** 0.207*** [0.123] [0.034] [0.142] [0.039]
Distance ‐‐ ‐‐ 0.013 0.003 [0.025] [0.007]
Adjacency ‐‐ ‐‐ 0.148* 0.041* [0.080] [0.022]
Homestatefixedeffects Regressiondiagnostics: Observations 10,224 10,224Pseudo‐R2 0.235 0.243p‐valuejointsignificanceofallvariables 0.000 0.000p‐valuejointsignificanceofallindividualvariables 0.000 0.000p‐valuejointsignificanceofparty‐dummies 0.000 0.000p‐valuejointsignificanceofstatecontrols 0.000 0.000p‐valuejointsignificanceofcrossstatevariables 0.000 0.000Notes:*/**/***denotesignificanceatthe10%/5%/1%level;Standarderrorsinbrackets;abasecategoryisthemarketorientedliberaldemocraticparty“FDP”;bStatecharacteristicsare2010dataforsurveywaves1and2,whichbothtookplacein2011,and2011dataforsurveywave3,whichtookplacein2012.Weightingbasedoninverseresponseprobabil‐itiesbasedonpartyandstateaffiliation.
top related