Role of bank credit on local growth: do politics and crisis matter?

(1)

ContentslistsavailableatSciVerseScienceDirect

Journal

of

Financial

Stability

journal homepage:www.elsevier.com/locate/jfstabil

Role

of

bank

credit

on

local

growth:

Do

politics

and

crisis

matter?

Zeynep

Önder

1

_,

_Süheyla

_{Özyıldırım}

∗

BilkentUniversity,FacultyofBusinessAdministrations,Ankara,Turkey

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received28September2011

Receivedinrevisedform13April2012

Accepted3December2012

Available online 17 December 2012

JELclassiﬁcation: G21 Keywords: State-ownedbanks Bankcredit Localgrowth Crisis Election

a

b

s

t

r

a

c

t

Althoughstate-ownedbanksareexpectedtopromotethegrowthofless-developedregions,especiallyin developingeconomies,severalcross-countrystudiesreportthatlendingbystatebanksisassociatedwith theinefficientallocationofcreditandlowlevelsofdevelopment.Further,statebankshavebeenfoundto lendtotheircronies,especiallyaroundelections.Inthispaper,westudythelendingactivitiesof state-ownedandprivatebanksduringtheperiod1992–2010andanalyzetherelationshipbetweenthecredit thesebanksprovideandlocaleconomicgrowthinTurkeyduringcrisisperiodsandinelectionyears.We findthattheshareofstate-ownedbanksinthecreditmarketincrisisperiodsandlocalelectionyearsis significantlyhigherthantheirshareinnon-crisisandnon-electionperiods.Thepercapitarealcreditthat state-ownedbanksprovideduringcrisisyearsisfoundtobepositivelyassociatedwithlocalgrowthin allprovinces.Ourresultssuggestthatalthoughstate-ownedbanksmightissueloansforpoliticalreasons inelectionperiods,theyalsoseemtoplayanimportantroleinoffsettingtheadverseeffectsofeconomic shocks,especiallyindevelopedprovinces.

1. Introduction

The 2007–2008 financial crisis restarted the long-standing debateonthe roleof state-owned banks.2 _Those_in _support_of state-ownedbanksclaimthatthebanksplayanimportantrole inexpandingaccesstofinancialservicesforsmallbusinesses3_or forsociallydesirableprojectsthatprivatebanksignore.Despite thetheoreticaladvantagesofthesebanksinbroadeningaccessto credit,theyaregenerallyregardedasinefficientandvulnerable topoliticalinterferences.Forexample,LaPortaetal.(2002)report

∗ Correspondingauthor.Tel.:+903122901899;fax:+903122664958.

E-mailaddresses:zonder@bilkent.edu.tr(Z.Önder),suheyla@bilkent.edu.tr(S.

Özyıldırım).

1 _Tel.:₊₉₀₃₁₂₂₉₀_2038;_fax:₊₉₀₃₁₂₂₆₆_4958.

2_For_the_most_recent_discussion_about_the_role_of_the_state-owned_banks,_see_the

blogmaintainedbyDemirguc-Kunt(http://blogs.worldbank.org/allaboutﬁnance/

the-aaf-virtual-debates-join-charles-calomiris-and-franklin-allen-in-a-debate-on-state-owned-banks).

3_Beck_and_{Demirguc-Kunt}₍₂₀₀₆₎_highlight_that_access_to_and_the_cost_of_ﬁnance

areoftenrankedasthemostconstrainingfeaturesofthebusinessenvironmentby

smallandmediumenterprises(SMEs)andthatstate-ownedbanksareexpected

tosolvetheirﬁnancingproblems.However,empiricalstudieshaveyieldedmixed

evidenceinfavorofstate-ownedbanks.Forexample,Clarkeetal.(2005)showthat

state-ownedbanksdonotnecessarilyoutperformprivatebankswithrespectto

smallbusinesslendinginLatinAmerica(Argentina,Chile,Colombia).Becketal.

(2011)documentthatthereisnosigniﬁcantassociationbetweengreaterstate

ownershipofbanksandﬁnancialaccessofSMEsacrossasampleof38

develop-ingcountries.Ontheotherhand,BurgessandPande(2005)ﬁndthata1%increase

inthenumberofruralbankinglocationsinIndiareducesruralpovertyby0.34%.

thatgovernmentownershipofbankscouldhaveharmfuleffectson productivity,economicgrowth,andthedevelopmentoffinancial systems.Miccoetal.(2007)documentthatstate-ownedbanksin developingcountrieshavelowerprofitabilityratiosthantheir pri-vatecounterparts,andthatthegapintheirperformancesincreases duringelectionyears.Cornettetal.(2010)findthatstate-owned banksaregenerallylessprofitable,havegreatercreditriskthan privatelyownedbanks,andfinancethegovernmenttoagreater degree.Severalotherstudiesexaminestate-ownedbanksina sin-glecountryandpresentevidencethatthesebanksprovidemore fundingorchargelowerinterestratestopoliticallyconnectedfirms orindividuals(forexample,seeSapienza(2004)forItaly,Leuzand Oberholzer-Gee(2006)forIndonesia,KhwajaandMian(2005)for Pakistan,andImai(2009)forJapan).

Whilemostofthepreviousstudiesreportanegativeor insignif-icanteffectofstate-ownedbanksongrowth,afewrecentpapers document their positive association with it. Andrianova et al. (2012)examinemorethan100countriesduring1995–2007and showthatgovernmentownershipofbankswasassociatedwith higherlong-rungrowthwhencontrollingfor regulatoryquality. Similarly,KörnerandSchnabel(2009)observethepositiveeffects ofstateownershipofbanksontheaveragegrowthrateofpercapita GDPincountrieswithwell-developedﬁnancialsystemsand better-qualitypoliticalinstitutions.ÖnderandÖzyıldırım(2010)analyze theeffectofstate-ownedbankcreditonlocalgrowthinTurkeyand showthattheloansprovidedbystate-ownedbanksareassociated withhighereconomicgrowthinthecountry’sdevelopedprovinces. However,thisrelationisnotobservedinTurkey’spoorprovinces,

(2)

whichsuggeststhattheroleofstate-ownedbanksmightdepend ontheexistenceofinstitutionalqualityevenatthemicrolevel.

Ingeneral,empiricalstudiesseemtohavenostandardanswer thatisvalidforallcountriesandforallsituationsregardingtherole ofstate-ownedbanks.Dothesebanksprovidefundingforpolitical reasonsor toreducemarket failures? Tocontribute this litera-ture,westudythelendingbehaviorofstate-ownedandprivate banksduringcrisisandelectionyearsandanalyzewhethercredit providedbythesebanksintheseyearsaffectsoutputgrowthat theprovinciallevelinTurkey.AshighlightedbyLevy-Yeyatietal. (2007),privatebanksmightoverreacttorecessionsandamplify thebusinesscycle.Thus,publicbankscanbeusedasacrisis reso-lutionvehicletodistributesubsidiestosomesectorsorregions. Obviously, regions areaffected by financialcrises with varying severity;developedareasmightbehitharderthanless-developed areas.4_However,_to_the_best_of_our_knowledge,_no_study_in_the lit-eratureexamineshowlendingbystate-ownedandprivatebanks variesduringcrisisperiods,whethertheloansprovidedby state-ownedbanksintheseperiodshaveanywelfareeffect,orwhether theseeffectsdifferfordevelopedandless-developedregions.Thus, ourfirstaiminthispaperistotestwhetherthesebankshaveany beneficialroleduringeconomicslowdowns.5

Severalempiricalstudieshaveexaminedthelendingactivities of state-owned banksduring electionyears. For example, Dinç (2005)analyzeslendingbybanksin36countriesandreportsan increaseinlendingbystate-ownedbanksduringelectionyears, mainlyinemergingmarkets.Similarly,Miccoetal.(2007)findthat state-owned banksindeveloping countriesincrease theirloans inelectionyears.Cole(2009)showsthatstate-ownedbankslend moreagriculturalcreditinelectionyearsinIndiabutthatthisextra credithasnosignificanteffectonagriculturalrevenueorcropyield atthedistrictlevel.Previousfindingssuggestthatpublicsources, includingstate-ownedbankfunds,areusedmoreforthe advance-mentofgovernments’ambitionsinelectionperiods.However,the importantquestioniswhetherfundsusedforpoliticalreasons ben-efitthewellbeingoflocalpeopleaswell.Thus,thesecondaimof thispaperistoprovidetheanswertothisquestion.Bycomparing thewelfareimplicationsofloansissuedbystate-ownedand pri-vatebanksonpercapitaprovincialincomeinelectionyears,we trytounderstandhowapoliticaleventmightchangelocal pros-peritywhenbothstate-ownedandprivatebankshaveaccesstothe samelocalinformation.

We study thelending activities of Turkishstate-owned and privatebanksduringtheperiod1992–2010.Turkeyisan interest-ingcountryinwhichtoanalyzetheassociationbetweenlending by thesebanksand localgrowthin crisisand electionperiods. First, over thesample period, theTurkish economy washit by threecrises,andstate-ownedbanksareconsideredtobeamong the prime contributors to one of these crises (Steinherr et al., 2004;Alper andOnis,2004).Second, thesebanksare themain providersof accesstoﬁnancing in thelaggingareasof Turkey.

4_The_more_modernized_areas_of_Indonesia_(Java_and_urban_areas)_were_hit

harderbytheAsiancrisisthanlessmodernizedoragriculture-dominatedareas

(Brodjonegoro,2002).Aftertherecentcrisisin2007–2008,disposableincome

decreasedmorethan5%inthenorthernprovincesofItalybutdecreasedonlyby

about1.4%inthesouthernpart(RizziandDallara,2011).

5_In_the_literature,_choosing_the_best_way_of_resolving_economic_shocks_and

acceleratingeconomicrecoveryisstilladebatedissue.Thereisevidencethat

gov-ernmentsmaycreatelargeﬁscaloutlaysinresolvingabankingcrisiswiththe

objectivetoacceleraterecovery.Forexample,LaevenandValencia(2008)

exam-ine42crisisepisodesduring1970–2007andpresentthatstate-ownedbanksmay

becausinglargeﬁscalcostsandencourageunproductiveriskstobetakenat

gov-ernmentexpense.Althoughgrowthandﬁscalissuesareintertwinedregardingthe

roleofstate-ownedbanks,duetodataconstraintsattheprovinciallevel,inthis

studyweonlyconcentrateonthegrowthissue.

Althoughseveralstate-ownedbankshavebeenprivatizedsince the1980s,threestate-ownedbanksarestillavailabletoallocate funds toinvestment projects in poor regions. Third,control of state-ownedbanksisveryimportantforpoliticalparties,which suggeststhat theymight providefundingfor politicalreasons.6 Hence,comparingtheeconomicsignificanceofstate-ownedand privatebanks,especiallyattheprovinciallevel,contributestothe literatureonunderstandingtheroleofinstitutionalframeworks forregionaldevelopment,mainlyduringcrisisorelectionperiods. Ourresultsindicatethatthecreditmarketshareofstate-owned banksincrisisandlocalelectionyearsissignificantlyhigherthan theirshareinnon-crisisandnon-electionyearsinallprovinces, andinless-developedprovinces.We findthat state-ownedand privatebankcreditpercapitaalikearepositivelyandsignificantly associatedwiththewell-beingofpeopleinallTurkishprovinces. Moreover,creditprovidedbystate-ownedbanksduringcrisisyears isfoundtohaveasignificantandpositiveeffectonlocalgrowthin allprovincesandindevelopedprovinces,butcreditgrantedduring electionyearsdoesnothaveasignificanteffect.Theseresults sug-gestthatalthoughstate-ownedbanksmightissueloansforpolitical reasonsduringelectionperiods,theymightalsoplayanimportant roleinoffsettingtheadverseeffectsofeconomicshocks,especially indevelopedregions.

Thepaper is organized as follows: Section2 provides infor-mation about banking in Turkey. Section3 summarizes the hypotheses.Section4presentstheempiricalmodelandthedata. Section5reportstheresultsandseveralrobustnesschecks. Sec-tion6concludesthepaper.

2. TheTurkishbankingsector

TheTurkishbankingsectorconsistsofdepositbanks, develop-mentandinvestmentbanks,andparticipationbanksthatoperate accordingtotheprofit/loss-sharingprinciple.In2010,ofthe48 banksintheTurkishbankingsector,31weredepositbanks,13were developmentandinvestmentbanks,andfourwereparticipation banks.Thenumberofstate-ownedbankswaseightin1990and threesince2001(Table1).Asinmostotherdevelopingcountries, state-ownedbanksinTurkeyoriginatedwithaclearmandateto financespecificsectorsoftheeconomy.7_They_held_31%_of_the_total assetsoftheTurkishbankingsectorattheendof2010.

State-ownedbankshavecreatedafairamountofprice distor-tionintheTurkish bankingsectorfor years.Theirrevenuesdid notreflectrealincome,asinterestearningsonuncollected“duty losses”accruedasincomeandwerereportedonthebalancesheet asreceivablesfromthetreasury.However,theycarriedouttheir tasksofextendingsubsidizedcreditinthemarketwithshort-term financing athigher costs. Eventually,thelosses of these banks made them an instability factor in the financial sector (BRSA, 2010).8_With_the_Turkish_banking_crisis_in_2001,_the_government realizedthatthecurrentpublicbankingpracticescouldnolonger continue.Inthesecondquarterof2001,thegovernmentfinancially andoperationallyrestructuredstatebanks:preferentialcreditas partofsupportprogramsceased,andstate-ownedbanksbecame

6 _For_example,_during_the_1997–1999_coalition_government_period,_control_of_the

existingpublicbankswasdistributedamongthreepartiesinthecoalitionaccording

totheirvotingshares.

7 _Ziraat_Bank_was_established_in₁₈₆₃_to_offer_mainly_agricultural_credit;_Emlak

Bankwasfoundedin1927toprovidemortgageloans;Etibank(1935)helpedthe

developmentofminingandpowersuppliesthroughcreditandHalkBank(1938)

hassuppliedspecializedcredittosmallbusinessesandcraftsmen.

8 _During_the_week_of_February_19,_2001,_the_two_largest_state-owned_banks,_Ziraat

BankandHalkBank,hadlossesthatamountedto$2.5billion,orabout2%oftheGDP

(3)

Table1

SomecharacteristicsofdepositbanksinTurkey,1990–2010.

State-ownedbanks Privatebanks

1990 2010 Meana _Domestic_banks _Foreign_banks

1990 2010 Mean 1990 2010 Mean Numberofbanks 8 3 4 25 11 24 23 17 17 Numberofbranches 2975 2744 2623 3455 4582 3684 113 2096 546 Numberofemployees 80,825 47,235 59,249 68,145 82,270 72,103 3012 42,013 12,106 Branch/bank 372 915 655 138 417 191 5 123 33 Employee/branch 27.2 17.2 22.2 19.7 18.3 19.6 26.7 20.0 26.9

Totalassets(USDmillion) 25,917 193,850 60,805 25,156 322,919 110,177 2011 88,171 20,003

Totalloans(USDmillion) 12,340 95,344 20,305 10,809 170,322 50,005 969 52,956 11,134

Totaldeposits(USDmillion) 15,797 148,465 45,530 11,337 200,329 69,252 776 50,956 11,497

Deposits/assets(%) 61.0 76.6 72.9 63.6 72.1 63.9 57.8 57.8 48.2

Soundnessmeasures(%)

Non-performingloans/loansb _1.7 _3.3 _11.4 _0.5 _3.3 _5.6 _0.3 _6.1 _3.9

Capital/totalassets 8.2 9.9 7.7 11.5 13.4 11.8 11.1 13.6 14.9

Returnonassets 2.2 2.3 1.5 3.5 2.4 2.8 3.6 1.4 4.7

Liquidassets/totalassets 29.9 30.2 42.6 39.0 34.1 42.3 42.5 33.9 49.7

Loans/totalassets 47.6 49.2 34.2 43.0 52.7 41.4 48.2 60.1 39.0

Source:TurkishBankingAssociation.

a_Mean_columns_report_average_values_for_the_period_1990–2010.

b_Loans_include_total_loans_and_receivables.

subjecttothesameregulationsandsupervisionsasprivatebanks.9 Asaresult,theyhavebecomeincreasingly“prudent”inaccepting badloans andwaivingappropriate provisions, andare creating more transparent balance sheets, and have more reasonable amountsforloanportfolios.Theirnon-performingloans-to-gross loansratiodecreasedfrom18.6%in2001to3.8%in2010;theirtotal loans-to-totalassets ratioand theirliquid assets-to-totalassets ratioincreasedto49.2%and30.2%,respectively,in2010(Table1). Privately owned banks have dominated the Turkish bank-ing sectorsince the early1990s. Someprivate domestic banks arelarge-scale,providingdiversebankingservicesthroughtheir nation-widebranchnetworks.Someprivatelyowned banksare small-ormid-sized,concentratingtheiractivitiesin larger cen-tresandengagingmainlyinwholesalebanking.Duringthe1980s and1990s,thegovernment’sborrowinghadabsorbedan impor-tantportionoftheloanablefundsintheeconomy,leavinglittle fortheprivatesector.However,thedeclineingovernment bor-rowing,inﬂation,andrealinterestratesafterthebankingcrisisin 2001allowedprivatebankstoconcentratemoreonlendingtothe privatesector.Themarketsharesofprivatedomesticbanksinthe creditmarketincreasedfrom39.5%in1990to53.5%in2010.10

ForeignbankentryintoTurkeybeganwithitsﬁnancial liber-alizationin the1980s.Ofthe43banksin1980,onlyfourwere foreign.Currently,therearemoreforeigncommercialbanks oper-atinginTurkeythandomesticbanks(Table1).However,theirshare wasverysmalluntilrecently.Theassetshareofforeignbankswas below5%in2005andincreasedto14.6%in2010.Similarly,their shareinthecreditmarketincreasedafter2005andbecame16.6%in 2010.Bytheacquisitionoffourdomesticallyownedprivatebanks, thenetworkofforeignbanksincreasedmorethanﬁvefoldduring theperiod2005–2010.

Asof2010,onaverage, depositsconstituted62%ofthetotal resourcesin the Turkish banking sector.Thisratio is over 70% forstate-ownedand privatedomestic banks.Assetquality,that

9_Until_a_new_autonomous _agency,_the_Banking _Regulation_and_Supervision

Agency(BRSA)wasestablishedinSeptember2001toregulatethebanking

sec-tor,bankingsupervisionhadbeensplitbetweentheCentralBank(off-site)andthe

Treasury(on-site).

10_In_2010, _commercial_banks_issued_96.8% _of_all_loans._The_remaining_3.2%

belongedtodevelopmentandinvestmentbanks.

is, non-performing loans as a fraction of gross loans, and the capitalization ratio of state-ownedand domestic privatebanks improved considerably during the last decade (Table 1). Yet, duetoincreaseincompetitioninthesector,especiallywiththe penetrationofforeignbanks,theproﬁtabilityofprivatedomestic bankshasdeclinedinrecentyears.

Comparedtothesizeoftheeconomy,theTurkishbanking sec-torisstillrelativelysmall.Theratiooftotalassetsofallbanksto GDPwas29%in1990andincreasedto91%attheendof2010.The credit/GDPratioandthedeposits/GDPratiowereonly15%in1990 andincreasedto43%and54%,respectivelyin2010.11_These_ratios arefarbelowtheEUaveragebutclosertotheratiosinthetransition economiesoftheEU.

The Turkish banking industry is dominated by seven large banks:threestate-ownedandfourprivatedomesticbanks.They held78.2%ofthesectorassetsinDecember2010.Theshareofthe ﬁvelargestbanks– acommonindicatorofconcentration–stands at62.9%andindicatesafairlevelofcompetitioninthesector.

Bank branchesareoneofthefundamentalchannelsfor pro-vidingbankingservicesinTurkey.Thereare9422branchesof31 deposit banksasof December2010and 29%of thesebranches belongtothreestate-ownedbanks.Thesevenlargestbankshave branchesinallofTurkey’s81provinces.Attheprovinciallevel, branchdensityislowcomparedtodevelopedcountriesbutalmost thesameastheworldaverage(InternationalFinanceCorporation, 2010).Therewere13branchesforevery100,000personsin2010. 3. Hypotheses

Inthispaper,weexaminethelendingactivitiesofstate-owned andprivatebanksincrisisyearsandingeneral(i.e.,parliamentary) andlocalelectionyears.First,weanalyzewhetherthesharesof state-ownedandprivatebanksincreditmarketsaredifferentin crisisandnon-crisisyearsandinelectionandnon-electionyears. Then, we study how the relationship between credit provided by banks withdifferentownership and local economic growth changesintheseyears.Weconducttheanalysesforallprovincesas wellasforasubsampleofdevelopedandless-developedprovinces.

11_GDP_in_current_prices_was_150,676_million_USD_in₁₉₉₀_and_became_735,264

(4)

3.1. Thesharesofstate-ownedandprivatebanksinthecredit market

3.1.1. Bankcreditduringcrisisyears

Duringperiodsofeconomiccrisis,banksreducetheirlending activitiesbyinvestinginsecurities,especiallygovernment secu-rities(BernankeandLown, 1991).However,somestudiesshow thatthelendingactivitiesofstate-ownedbanksareless respon-sivetomacroeconomicshocksthanthoseofprivatebanks(see, forexample,DetragiacheandGupta(2006)forMalaysiaandMicco andPanizza(2006)fordevelopingcountries).MiccoandPanizza (2006)listfourreasonsfortheinsigniﬁcantchangeinthe behav-iorof state-owned banksin crisisyears:(i)state-owned banks trytostabilizethecredit marketaspartof governmentpolicy, (ii)theyhaveexcessloanablefundsbecausedepositorspreferto channeltheirsavingstothesebanksassafehavensduring reces-sions, (iii)themanagers ofthesebanksmight beslowtoreact toshocks,and(iv)politiciansmightinﬂuencethelending behav-iortoincrease theirprobabilityofre-election.Inthis paper,we postulate that state-ownedbanksdo not reactasmuch as pri-vatebanksineconomiccrisisperiodsandexaminethishypothesis at the provincial level by using a simple market share mea-sure.

3.1.2. Bankcreditduringelectionyears

Electionsareone ofthemoststudiedpolitical eventsin the bankingliterature,andseveralstudieshavedocumentedthat state-ownedbankschangetheirlendingbehaviorduringelectionyears. Dinç(2005)examinesthebehaviorofthetenlargestbanksin17 developedand19developingcountriesandfindsthatstate-owned banks(relativetoprivatebanks) increasetheirlendingin elec-tionyearsonlyindevelopingcountries.MiccoandPanizza(2006) alsoobservelendingduringelectionyearsbutfindnosignificant relationbetweenloangrowthandpoliticallendinginindustrial ordevelopingcountries.Later,Miccoetal.(2007)findthat state-owned banks perform worse than private banks and that the disparitybetweenthemwidensduringelectionyears,especiallyin developingcountries.12_Baum_et_al.₍₂₀₁₀₎_examine_the_effects_of generalelectionsontheTurkishbankingsystembetween1963and 2007andobservenosignificantdifferencebetweenlending behav-iorofstate-ownedandothertypesofbanksduringparliamentary elections.Jackowiczaetal.(2012)findthatalthoughstate-owned bankschargesignificantlylowerinterestratesonloansintheyears ofparliamentaryelectionsinCentralEuropeancountries,theirloan volumedoesnotchangesignificantly.

In this paper, we examine local and general elections. We hypothesizethatstate-ownedbanks’shareof thecredit market inelectionperiodsissigniﬁcantlygreaterthantheirsharein non-electionperiods.Since1986,provincialmunicipalitieshavebeen allowedbylawtoborrowfromalltypesofcommercialbanks(i.e., state-owned,private,or foreignbanks).Hence,it wouldnot be surprisingtoobservea largerprovisionbyprivatebanksinthe creditmarketinelectionyears;however,theirshareisexpectedto declineintheseperiods.

3.1.3. Bankcreditinless-developedanddevelopedprovinces Increasinggrowthandreducing economicdisparitybetween regionsaremajorchallengesforincumbentgovernmentsinmany developingeconomies.TheTurkishgovernmenthasbeen classi-fying certainprovincesas“priority provincesfor development”

12_They_deﬁne_‘election_years’_as_years_with_presidential_elections_in_countries

withapresidentialsystemandlegislativeelectionsincountrieswithparliamentary

systems.Theirsamplecomprises179countries.

since1968toprovideincentivesandsubsidiestoimprovegrowth in these provinces. The number of provinces included in this list has increased over time and now stands at 50 out of 81 provinces.In this paper, we use this classiﬁcation to examine whetherstate-ownedbanksbehavedifferentlyindeveloped (non-priority) and less-developed (priority) provinces in crisis and electionyears.

InTurkey,ÖnderandÖzyıldırım(2010)findthatcreditprovided bystate-ownedbankshasanegativeorinsignificantassociation withlocalgrowthin“priority”provincesbutapositive associa-tionwithgrowthindevelopedprovincesfortheperiod1992–2000. Inanotherstudy,theyshowthatcreditprovidedbystate-owned banks is associated with local growth in “priority” provinces if they are politically connected with the government (Önder andÖzyıldırım, 2011).Based onthesefindings, we expectthat state-ownedbankswillbehavedifferentlyindevelopedand less-developedprovincesinthecrisisandelectionyears.

3.2. Therelationshipbetweenbankcreditandlocalgrowth

Afterexaminingwhetherthecredit marketshare ofTurkish state-ownedbanksishigherinelectionandcrisisyears,wetest whetherthepercapitacredittheyprovideduringtheseperiods is positivelyassociated with per capitaGDPgrowth ratein all provincesregardlessof theirdevelopmentlevel, controllingfor otherfactors that might affectgrowth rate. Although previous empiricalstudiesshowthatgenerallythemanagementof state-ownedbankshaspoliticalincentivesinallocatingcreditregardless of thestate of theeconomy (see for example,Sapienza, 2004; Imai, 2009), wedo nothave an a priori expectation aboutthe contributionofthesebanksoneconomicgrowthinnormaltimes. However,wemayobserveapositiveassociationbetweencredit providedbystate-ownedbanksandprovincialgrowthduring cri-sisperiodsifstate-ownedbanksfillfinancialgapspossiblycreated byprivatebanksandfinanceeconomicallydesirableprojectsthat areignoredbyprivatebanks.Nevertheless,governmentsmayalso makepoliticallymotivatedlendingdecisionsduringcrisisperiods toimplementtheirdiscretionarypolicies(seeBonginietal.(2001) duringthe1997–1999EastAsiancrisisandMcQuerry(2001) dur-ingthe1994Brazilian bankingcrisis).Ifstate-ownedbanksare morelenientandlessprudentduringcrisisperiods,noassociation withgrowthmaybeobserved.

Similarly,wemayseeloanexpansionbyTurkishstate-owned banksduringelectionperiods,butlenient lendingbehaviorand marginallygrantedcreditmayincreasedefaultratesinthewakeof elections.Hence,likenormalperiods,wehavenoapriori expecta-tionontherelationshipbetweenlocalgrowthandloansprovided by state-owned and private banksduring periods of increased uncertaintyandasymmetricinformation.

4. Empiricalmodelanddata 4.1. Themodel

Toexaminetherelationshipbetweenbankcreditandeconomic growthattheprovinciallevelovertheperiod1992–2010,we esti-matethefollowingﬁxed-effectsmodel:

Yit=ˇ0+ˇ1BankCreditit+ˇ2PublicExpenditureit

+ˇ3Urbanizationit+ˇ4Schoolingit+ˇ5GDPit−1+ˇ6DCrisis

+ˇ7DElection+ˇ8Dj×BankCreditit+

i+

it

whereYitisthegrowthrateofrealGDPpercapitainprovincei

(5)

creditorpercapitaprivatebankcreditprovidedinprovinceiat timet.13

Wetrytocontrolforotherfactorsthatmayaffectaprovince’s percapitaincomegrowthrate.Thecontrolvariables arepublic expendituresper capita(PublicExpenditure),theshareof urban population(Urbanization),a measure of humancapital (School-ing),andthelaggedpercapitarealoutput(GDP_t−1).Weexpectthe coefﬁcientonpublicexpendituresandurbanizationtobepositive andthecoefﬁcientonincomeleveltobenegative.Human capi-talrepresentsthenumberofhigh-schoolstudentsperhigh-school teacher,andtheexpectation isthat thisvariable hasa positive relationtoprovincialgrowthrate.DCrisisandDElectionaredummy

variablesindicatingcrisisandelectionyearsrespectively.

isa vectorrepresentingprovincialﬁxed-effects.

We examine howthe relationshipbetweenbank credit and growthchangesincrisisorelectionyearsbytheinteraction vari-ables(Dj×BankCreditit,wherej=Crisis,Election(LocalorGeneral)).

ˇ8measuresthechangeinthecoefﬁcientonpercapitacredit

pro-videdbyeitherstate-ownedorprivatebanksincrisisorelection years.

Turkeyexperienced three major economic crises during the sampleperiod.Thefirstcrisiswasofshort-duration,in1994,and wasa currencycrashand a bankingcrisis.Threeprivatebanks failed. Fiscal fundamentals deteriorated significantly. The huge domesticcreditexpansiontothepublicsectoreventuallycaused drasticdepreciationoftheTurkishLira (TL)andresultedin the lossofforeignassetsoftheCentralBankoftheRepublicofTurkey (CBRT).Afteranoutputlossof6.1%in1994,outputrecoverywas rapid.Theeconomygrew7.5%and8%in1995and1996, respec-tively.Thesecondcrisishadalong-duration,spreadingoverthree years(1999–2001).In1999,therewasanothercurrencycrashand through2000and2001anotherbankingcrisis.TherealGDPgrowth ratewas−3.37%in1999,6.77%in2000,and−5.67%in2001.Turkey implementedanIMF-supportedprogramatthebeginningof2000, whichsucceededtosomeextent,reversingthenegativetrendin GDPfor thatyear.However,risingcurrentaccountdeficits, real appreciationoftheTL,andthefragilityofstate-ownedand pri-vatebankswereconsideredthecauseoftheseverefinancialcrisis in2001,sotheprogramultimatelyfailed.Theeffectsofthis sec-ondcrisisonthebankingsectorweremoreseverethanthoseof thefirstone.Seventeenbanksfailedover three years.We con-sidertheglobalcrisisperiodsof2008and2009asTurkey’sthird crisisperiodbecauseitwasoneofthehardest-hitcountries.The recessionof2008–2009ledtoamassivecollapseinexportsand thesubsequentcontractioninGDPof14.7%duringthefirst quar-ter of 2009 (Alp and Elekdag,2011).By the second quarterof 2009,theeconomyhadquicklyrebounded.Financialmarketsseem toberesilientincontrasttopreviouscrises.Thus,inthe empir-icalmodel, the dummyvariable for crisis,DCrisis,takes a value

ofonein1994,1999,2000,2001,2008,and2009and0 other-wise.

Duringthesampleperiod,therewerefourlocalelections,in 1994,1999,2004,and2009.Hence,thedummyvariableDLElection

deﬁneslocalelectionperiodsandequalsoneintheseyearsand0 otherwise.Intheanalysisofgeneralelections,thedummyvariable, DGElectiontakesavalueofoneinthegeneralelectionyears(1995,

1999,2002,and2007)and0otherwise.14

Whenexaminingtheimpactofa crisis,we controlfor local election years rather than general election years because the

13 _Because_of_the_{multicollinearity}_problem,_we_did_not_jointly_estimate_per_capita

creditissuedbystate-ownedandprivatebanks.

14 _In_order_to_{differentiate}_the_pure_effect_of_crisis_or_election_years,_if_two_events

concurrentlyoccurinthesameyear,thenthatyearisexcludedfromtheestimations.

estimationsaredoneattheprovincialratherthannationallevel. Moreover,politicalpartiesseemtoviewlocalelectionsasmore crucialinexpandingtheirelectoralpopularityinthenational polit-icalarenabecauseofthelargeconcentrationsofvotersinurban areas.Forexample,Incioglu(2002)arguesthatinTurkey“...[t]he growingimportanceoflocalelections,especiallythoseconcerning municipaladministrations,isfurtherunderscoredbythefactthat themayorsoflargemetropoliseshavebecomemoreinﬂuentialin nationalpoliticsthanmanylegislatorsservingintheparliament.”

We divide theprovincesinto developed and less-developed accordingtothelistofpriorityprovincesthatCouncilof Minis-tersdeterminesannually.Themodel isestimatedforthesetwo subsamplesofprovinces.Weexpectthatstate-ownedbankcredit hasapositiveeffectonlocalgrowthinelectionyearsespeciallyin theless-developedprovincesbecausethepreviousﬁndings indi-catethat state-ownedbanksdo much betterinless-developed, politicallyconnectedprovinces.Moreover,weexpectthatan eco-nomicshockwillhaveagreaterimpactondevelopedprovinces. Hence,theextracreditthatastate-ownedbankprovidesduring crisisyearsisexpectedtohavemoreofaneffectonthegrowth rateindevelopedprovinces.

Inthegrowthliterature,creditisusuallytreatedasan endoge-nousvariable.Firstwetestwhetherstate-ownedandprivatebank creditisendogenousornot.Accordingtotheresultsofthe Haus-mantest,wecannotrejectthehypothesisthatstate-ownedbank creditisexogenousbutwecanrejectthehypothesisthatprivate bankcreditisexogenous.Therefore,weusetheOLSestimationin ouranalysisonstate-ownedbankcredit.Forestimationswith pri-vatebankcredit,weidentifytheshareofprivatebankbranches locatedinthatprovinceasanappropriateinstrumentalvariable (IV)15_and_estimate_our_models_accordingly.16

4.2. Data

Weconstructedanunbalancedpaneldataset17_by_employing annualdataoncreditprovidedbystate-ownedandprivatebanks intheprovincesbetween1992and2010.Alldataonbankcredit comefromtheCBRT.18_Provincial_public_expenditures_come_from theTurkishMinistryofFinanceandtheothervariablescomefrom theTurkishStatisticsInstitute.

Table2showsthemeanvaluesofbankcreditandprovincial characteristicsforthewholesampleandforthesub-samplesthat compriselocalandgeneralelectionsandnon-electionyears,crisis andnon-crisisyears,andless-developedanddevelopedprovinces. Duringthesampleperiod,theaveragegrowthrateofpercapita

15_Although_dynamic_GMM_is_widely_used_in_the_growth_literature_(for

exam-ple, seeBecketal.,2000; Levineetal.,2000),werejectedthehypothesesof

over-identiﬁcationrestrictionswiththeSargantestandthesecond-orderserial

correlationoferrortermswiththeArellano–Bondtest.Theestimateswouldnot

berobustusingthedynamicGMMapproach;therefore,weusedtheOLSandIV

methods.

16_The_correlation_coefﬁcient_between_the_provincial_growth_rate_and_the_share_of

privatebankbranchesis−0.048whereasthecorrelationcoefﬁcientbetweencredit

issuedbyprivatebanksandtheshareofprivatebankbranchesis0.762.

17_The_number_of_provinces_changed_during_the_sample_period_from₆₇_in₁₉₉₂

to81in2000.Fourteennewprovinceswereestablishedfromdistrictsofexisting

provinces.Intheestimations,formerprovinceswereexcludedfromthesamplein

theyearwhennewprovincesweredelineatedfromthembecauseoftheartiﬁcial

declineinGDPleveloftheformerprovincesinthatyear.

18_In_the_empirical_analysis,_we_use_only_cash_loans_provided_by_the_{corresponding}

banksascredit.Inadditiontocashloans,bankscanissuenon-cashloans,suchas

lettersofguarantee,acceptances,andcredit.

TheCBRTclassiﬁesbanksasstate-ownedorprivatelyowned.Inthisgrouping,

for-eignandparticipationbanksareconsideredprivatebankswhenprovidingcredit

dataattheprovincelevel.Becauseofdataavailability,weusetheCBRT’s

(6)

Table2

Descriptivestatisticsofvariables.

All Crisisyears Electionyears Provinces

No Yes Local General Developed Lessdeveloped

No Yes No Yes

PercapitarealGDPgrowthrate(%) 2.89 5.86 −3.42 3.85 −0.66 2.91 2.83 2.44 3.17

(7.28) (5.38) (6.75) (6.60) (8.49) (7.33) (7.09) (6.70) (7.60)

Bankcreditpercapita(TL)

State-ownedbanks 502.43 469.79 571.75 493.31 536.14 517.47 446.87 645.73 414.44

(656.37) (634.6) (695.98) (657.4) (652.46) (659.49) (642.69) (628.84) (657.74)

Privatebanks 795.54 775.67 837.74 802.5 769.86 820.87 701.99 1531.39 343.7

(1525.51) (1474.45) (1629.43) (1540.59) (1470.53) (1553.12) (1417.22) (2207.92) (478.63) Publicexpenditurespercapita(TL) 134.34 138.11 126.31 136.42 126.65 141.48 107.92 143.1 128.95

(154.98) (163.42) (135.13) (158.25) (142.23) (160.62) (128.9) (204.38) (114.28)

Urbanization(%) 56.69 56.33 57.44 56.50 57.35 56.72 56.56 62.98 52.81

(13.32) (13.17) (13.62) (13.22) (13.66) (13.45) (12.84) (14.83) (10.58) Schooling

Studentsperteacher 17.56 17.48 17.71 17.57 17.49 17.51 17.72 16.70 18.11

(5.31) (5.44) (5.02) (5.45) (4.76) (5.27) (5.42) (4.08) (5.90)

DCrisis(%) 31.99 0.00 100.00 20.49 74.53 33.84 25.16 31.81 32.11

DLElections(%) 21.29 7.97 49.58 0.00 100.00 20.24 25.16 21.27 21.30

DGElections(%) 21.29 23.43 16.74 20.24 25.16 0.00 100.00 21.27 21.30

Note:Crisis(DCrisis)yearsare1994,1999,2000,2001,2008,and2009.Localelection(DLElections)yearsare1994,1999,2004and2009.Generalelection(DGElections)yearsare

1995,1999,2002,and2007.Standarddeviationsarereportedinparentheses.Monetaryvaluesareexpressedintermsof2010prices.

provincial GDP was 2.89%. It changes signiﬁcantly in the

sub-samples.Forexample,itwas−0.66%duringlocalelectionyears

and −3.42%duringcrisisyears. Less-developedprovinceshada

highergrowthratethandevelopedprovincesduringtheperiod

1992–2010(3.17%versus2.44%).Partofthisgrowthratecanbe

explainedbythepeoplemigratingfromless-developedto

devel-opedprovinces.

Onaverage,state-ownedbanksprovided502.43TL19_per_capita

peryearinaprovince,whereasprivatebanksprovided795.54TL. Fig.1showshowpercapitaprovincialbankcreditchangedthrough thesampleperiod.Intermsof2010values,state-ownedandprivate banksseemtoprovidealmostthesameamountofcreditpercapita during 1992–2001.After2001,theamountof credit percapita frombothtypesofbanksincreasesexponentially.Moreover, pri-vatebankshaveahighergrowthratethanstate-ownedbanks.The highaverageincrisisyears,asreportedinTable2,canbeexplained bytheincreaseinthepercapitabankcreditinthelateryearsthat coincidedwiththecrisis.

Table2showsthatbothtypesofbankcreditpercapitaarehigher indevelopedprovinces.Privatebanksprovideonaveragealmost 4.5timesmorecreditpercapitaindevelopedprovincesthanin less-developedprovinces.Theaveragepercapitapublicexpendituresin aprovinceisalmostone-thirdofthemeanofpercapitacredit pro-videdbystate-ownedbanks.Moreover,averagepercapitapublic expendituresarelowerincrisisandelectionperiods.Considering thenatureofpublicexpenditures(infrastructure,health, educa-tion,civilservantsalaries,andtransferpayments),thetrendduring crisisperiodsislogical;however,alowermeanvalueduring elec-tionperiodsisunexpected.

5. Empiricalresults

5.1. State-ownedandprivatebanks’creditmarketshares

Totestthehypothesesabouttheequalityofstate-ownedand privatebanklendingactivitiesindifferentyearsandindifferent

19_All_monetary_values_are_expressed_as_of_their_values_at_the_end_of_2010._At_that

time,theexchangeratewas1USD=1.54TL.

provinces,wecalculatethecreditmarketsharesofbothbanktypes ineachprovinceandineachyear.Table3showsthemeanvalue ofthemarketsharesofeachtypeofbankinalternativeyearsand provincesaswellasthet-statisticsthattesttheequalityofthese meanvalues.Theresultsshowthatattheprovinciallevel, state-ownedbankshaveasigniﬁcantlyhighermarketshareincrisisyears (54.3%)thaninnon-crisisperiods(47.5%).20_On_the_other_hand,_the marketshareofprivatebanksinthecreditmarketissigniﬁcantly lowerincrisisyears(45.7%)thaninnon-crisisyears(52.2%).The CBRT(2010)alsodocumentstheassuranceofstate-ownedbank lendingintherecentcrisis.

Weobservesimilarbehaviorforstate-ownedbanksinlocal elec-tionyears.Wefindthatattheprovinciallevel,thesebankshave asignificantlyhighermarketshareinlocalelectionyears(52.2% versus49.0%).Ontheotherhand,privatebankshaveasignificantly lowermarketshare inlocalelectionyears thaninnon-election years.Althoughstate-ownedbankshaveahighermarketsharein generalelectionyears,nosignificantdifferenceexistsbetween gen-eralelectionyearsandnon-electionyears.Thisfindingsupportsthe resultsofBaumetal.(2010),whichholdthatstate-ownedbanksin Turkeydonotchangetheirlendingactivityduringgeneralelection years.

Whenwecompare thecredit marketsharesofbanksin dif-ferentprovinces,state-owned bankshave a significantlyhigher shareincreditmarketsofless-developedprovincesthaninthose ofdevelopedprovinces.Theaverageshareofstate-ownedbanks increasesto62.9%inless-developedprovincesanddecreasesto 40.4%indevelopedprovincesincrisisyears.Theirshareisaround 60%inlocalandgeneralelectionyearsalike.Privatebankshavea significantlylowermarketshareincrisisandlocalelectionyears. Thedifferenceinmarketshareissignificantinallprovincesand indeveloped provinceswhencomparinglocalelectionyearsto non-electionyears.

Knowingthat state-ownedbankshave a signiﬁcantlyhigher shareofthecreditmarketduringlocalelectionandcrisisyears,

20 _Although_the_average_market_shares_of_public_and_private_banks_are_around_50%,

itchangedsigniﬁcantlyovertime.Forexample,in1992,theshareofstate-owned

(7)

1992 1993 1994 199519961997 1998 1999 2000 2001 2002 2003200420052006 2007 2008 20092010 0.000 0.500 1.000 1.500 2.000 2.500 0.133 0.154 0.234 0.246 0.134 0.426 0.426 0.403 0.460 0.270 0.195 0.259 0.414 0.557 0.673 0.920 0.966 1.071 1.411 0.163 0.247 0.229 0.267 0.263 0.558 0.417 0.404 0.452 0.353 0.375 0.486 0.693 1.110 1.416 1.720 1.815 1.714 2.088 State-owned Banks Private Banks ('000 TL)

Fig.1.Provincialrealcreditpercapita,1992–2010.

Table3

MeansandstandarddeviationsofcreditmarketsharesofState-ownedandprivatebanksinTurkishprovincesindifferentperiodsandt-statisticsforequalityofmeans.

All Less-developedprovinces Developedprovinces

State-ownedbanks Privatebanks N State-ownedbanks Privatebanks N State-ownedbanks Privatebanks N Crisis No 47.52% 52.17% 1021 55.14% 44.35% 632 35.15% 64.88% 389 (22.76) (22.64) (22.16) (21.81) (17.74) (17.69) Yes 54.31% 45.70% 479 62.89% 37.13% 296 40.41% 69.56% 183 (22.89) (22.93) (20.50) (20.57) (19.53) (19.56) t-Stat −5.37*** 5.14*** −5.08*** 4.79*** −3.21*** 3.24*** Localelection No 49.01% 50.73% 1182 56.93% 42.63% 732 36.12% 63.91% 450 (23.04) (22.92) (22.02) (21.68) (18.37) (18.33) Yes 52.22% 47.79% 318 60.16% 39.87% 196 39.47% 60.50% 122 (22.73) (22.84) (21.48) (21.58) (18.71) (18.76) t-Stat −2.21** 2.04** −1.83* 1.58 −1.78* 1.81* Generalelection No 49.44% 50.30% 1181 57.11% 42.46% 732 36.93% 63.08% 449 (22.77) (22.65) (21.86) (21.52) (18.24) (18.22) Yes 50.61% 49.39% 319 59.49% 40.50% 196 36.46% 63.54% 123 (23.91) (23.95) (22.15) (22.24) (19.41) (19.39) t-Stat −0.81 0.63 −1.35 1.12 0.25 −0.25

Note:t-Statteststheequalityofmarketsharesofstate-ownedbanksandprivatebanksincrisisandnon-crisis(electionandnon-election)years.Standarddeviationsare reportedinparentheses.Themarketsharesofbanksattheprovinciallevelchangebetween0%and100%.*,**,and***indicatestatisticalsigniﬁcanceatthe10percent,5 percentand1percentlevels,respectively.

wefurtherexaminewhetherthiscredithasapositiveassociation

withindividuals’well-beingattheprovinciallevel.

5.2. Bankcreditandlocalgrowth

Table4 reports theresults of the fixed-effects model with-outinteractionvariables.Thebasicmodelwithstate-ownedbank creditexplains33.2%ofthevariabilityinthepercapitaGDPgrowth ratesintheTurkishprovincesandthemodelwithprivatebank creditexplains33.7%ofthevariability.Theprovincialfixed-effects arefoundtobesignificant.Therefore,theyareincludedinall esti-mations.

Percapitastate-ownedandprivatebankcreditarebothfound tobesignificantlyand positivelyassociatedwiththeprovincial percapitagrowthrateofrealGDP,controllingforotherfactors thatmightaffectprovincialgrowth,suchaspublicexpenditures, urbanization, schooling, and initialincome level of a province. Interestingly,thecoefficientonpercapitastate-ownedbankcredit issignificantindevelopedprovinces,whereasthecoefficientonper capitaprivatebankcreditissignificantinless-developedprovinces.

However,bothtypesofbankcredithaveapositiveandsignificant effectonlocalgrowthinallprovincesregardlessoftheir develop-mentlevel.Thisresultisslightlydifferentfrompreviousstudies that examine the role of state-owned banks in Turkey (Önder and Özyıldırım,2010; Önder and Özyıldırım,2011), which find aninsignificantcoefficientonpercapitacreditforallprovinces. Differentfromthosestudies,thisstudycoversaperiodwhenan independentbankregulatoryagencyexistsinTurkey.These find-ingsseemtosupporttherequirementsofferedbyGuidottietal. (2004) thatregulatoryinstitutionsshouldbeindependentfrom politicalinfluencesforthereliableoperationofthefinancial sys-tem.

Thepositivecoefﬁcientfoundonstate-ownedbankcrediteven indevelopedprovincesseemstosupportthedevelopmentview explanationofgovernmentownershipofbanks.21_However,_it_is

21_Proponents_of_the_development_view_claim_that_public_banks_cure_ﬁnancial

mar-ketfailuresandthereforeenhancesocialwelfareorstimulategrowth(see,e.g.,

(8)

Table4

Fixed-effectsmodel– bankcreditandlocalgrowth.

Allprovinces Less-developed Developed Allprovinces Less-developed Developed

BankCredit(ˇ1) 0.0089** 0.0049 0.0126** 0.0079* 0.0224*** 0.0027 (0.0038) (0.0051) (0.0054) (0.0044) (0.0075) (0.0051) PublicExpenditures(ˇ2) 0.0643*** 0.0921*** 0.0099 0.0777*** 0.1087*** 0.0150 (0.0188) (0.0271) (0.0249) (0.0192) (0.0268) (0.0264) Urbanization(ˇ3) 0.3399*** 0.5860*** 0.3047*** 0.3641*** 0.5035*** 0.3488*** (0.0430) (0.0704) (0.0574) (0.0504) (0.0758) (0.0691) Schooling(ˇ4) 0.0261*** 0.0273*** −0.0068 0.0226*** 0.0258*** −0.0080 (0.0077) (0.0094) (0.0136) (0.0077) (0.0093) (0.0139) GDPt−1(ˇ5) −0.0741*** −0.0819*** −0.1172*** −0.0821*** −0.1147*** −0.1100*** (0.0125) (0.0155) (0.0222) (0.0137) (0.0193) (0.0222) DCrisis(ˇ6) −0.0799*** −0.0823*** −0.0723*** −0.0793*** −0.0802*** −0.0739*** (0.0039) (0.0052) (0.0058) (0.0040) (0.0052) (0.0058) DLElection(ˇ7) 0.0424*** 0.0425*** 0.0376*** 0.0425*** 0.0474*** 0.0346*** (0.0062) (0.0082) (0.0091) (0.0063) (0.0084) (0.0092) Intercept(ˇ0) 0.7954*** 0.6499*** 1.5357*** 0.9452*** 1.1222*** 1.4442*** (0.1630) (0.1799) (0.2957) (0.1873) (0.2455) (0.3022) AdjustedR2 _0.3320 _0.3704 _0.3250 _0.3367 _0.3775 _0.3266 N 1225 747 478 1225 747 478

Note:Thistablepresentstheestimationofthebasicmodelwithprovincialﬁxed-effects.Themodelswithstate-ownedbankcreditareestimatedwithOLS.Becauseofthe

endogenityproblem,anIVmethodisusedintheestimationsofmodelswithprivatebankcredit.

*_Statistical_{signiﬁcance}_at_the_10%_level.

**_Statistical_{signiﬁcance}_at_the_5%_level.

***_Statistical_{signiﬁcance}_at_the_1%_level.

still surprisingthat credit fromprivatebanksin thedeveloped provincesisnotfoundtobesigniﬁcantlyassociatedwiththe wel-fareoftheseprovinces.

Table 3 shows that state-owned banks have a significantly highershareinless-developedprovinces.Theinsignificant coef-ficientonpercapitacreditfromthesebanksintheless-developed provincessuggeststhatthiscreditmightbeusedtofinancesocial projectsortofixmarketfailuresintheseprovinces.Analternative explanationmightbethepoliticaluseofstate-ownedbankloans. However,per capitacreditissued byprivatebanksand foreign bankshasasignificantandpositiveassociationwithindividuals’ percapitaincomeinless-developedregions.Thisfindingindicates thatprivatebanksmighthavebettertechnologiestoidentify pro-ductive projectsinthoseregions andmaybettermonitorthem toyieldmorereturntolocalinvestors,supportingtheresultfrom DelaTorreetal.(2010).Theyshowthatalltypesofbanks(large, small,domestic,andforeign)expandtheiroperationsintheSME segmentandcompetewithstate-ownedbanksaggressively.They arguethatprivatebanks,includingforeignbanks,have compara-tiveadvantagesinofferingawiderangeofproductsandservices throughtheuseofnewtechnologiesformonitoringandmanaging risks.

Similartootherstudiesintheliterature,publicexpendituresper capitaandurbanizationhaveapositiveassociationwiththelocal growthrate.Thepositivecoefficientonpublicexpendituresis sig-nificantinallandless-developedprovinces.Consideringthetypes ofpublicexpendituresmadeintheseprovinces,suchas infrastruc-ture,education,andhealth,itisnotsurprisingtoseethiseffect. Schoolinghasapositiveandsignificanteffectonlocalgrowthfor allandless-developedprovinces.Becausethisvariableismeasured bythenumberofhighschoolstudentsperhighschoolteachers, anincreaseinitmightresultinadeclineinthequalityof educa-tion.However,consideringanaverage17.6highschoolstudents perteacher,itcanbearguedthatanincreaseinthisnumberalso indicatesanincreaseinhumancapital,whichwouldhaveapositive effectonthegrowthofpercapitaprovincialincome.Asexpected, thegrowthrateinpercapitaGDPdeclinessignificantlyincrisis years.Controllingforbankcreditandpublicexpenditures,years withlocalelectionshaveahighergrowthratethanotheryears.

5.3. Bankcreditandlocalgrowthincrisisyears

Table5reportstheresultsofthefixed-effectsmodelwiththe crisisinteractionvariable.Percapitastate-ownedbankcreditis foundtobepositivelyandsignificantlyrelatedtothegrowthrate ofrealper capitaGDPin theTurkish provinces,controllingfor otherfactorsthatmightaffectprovincialgrowth,suchaspublic expenditures,urbanization,schooling,andaprovince’sinitialreal GDP.Itholdsforallprovincesaswellasthedevelopedand less-developedprovinces;however,thecoefficientisonlysignificant whenallprovincesareincludedintheanalysis.

Thecoefficientontheinteractionvariablebetweenpercapita state-ownedbankcreditandthecrisisdummyvariableis signifi-cantinthedevelopedprovinces.Whenthesetwocoefficientsare added,percapitastate-ownedbankcreditisfoundtohavea pos-itiveandsignificantrelationshipwithlocalgrowthincrisisyears. Thisresultsuggeststhatpercapitaloansissuedbystate-owned banksincrisisyearshelpincreasethegrowthlevelinthe devel-opedTurkishprovincesaswellasinallprovinces.22_A_1%_increase inpercapitastate-ownedbankcreditindevelopedprovincesin crisisyearsisexpectedtoincreasethegrowthrateofpercapita realGDPintheseprovincesby4.2%.Ontheotherhand,asimilar increaseinpercapitastate-ownedbankcreditissuedinnon-crisis yearsisfoundtoincreasethepercapitarealGDPgrowthrateby only0.9%.However,percapitastate-ownedbankcreditincrisis yearsisnotfoundtohaveasignificanteffectonlocalgrowthin less-developedprovinces.

The results are slightly different for private bank credit. Althoughpercapitacreditfromthesebankshasapositive associa-tionwithlocalgrowth,theonlysigniﬁcantandpositivecoefﬁcient isfoundintheless-developedprovinces,whereprivatebankshave

22 _By_slightly_changing_the_model_and_using_the_provincial_per_capita_credit_growth

rateofstate-ownedbanksratherthanthelevelofpercapitabankcreditasan

inde-pendentvariable,wetrytoestimatewhetherTurkishstate-ownedbanksplayeda

smoothingroleincrisisperiods.Weﬁndthatthegrowthrateofstate-ownedbank

creditpercapitaispositivelyandsigniﬁcantlyassociatedwiththegrowthrateof

(9)

Table5

Bankcreditandlocalgrowthincrisisyears.

BankCredit(ˇ1) 0.0086** 0.0074 0.0088 0.0066 0.0179** 0.0019

(0.0041) (0.0058) (0.0055) (0.0044) (0.0077) (0.0052)

DCrisis×BankCredit(ˇ8) 0.0010 −0.0063 0.0334*** −0.0819*** 0.0255*** −0.1085***

(0.0055) (0.0067) (0.0102) (0.0137) (0.0088) (0.0223) PublicExpenditures(ˇ2) 0.0645*** 0.0906*** 0.0170 0.3657*** 0.1052*** 0.3485*** (0.0188) (0.0272) (0.0248) (0.0503) (0.0266) (0.0691) Urbanization(ˇ3) 0.3404*** 0.5771*** 0.3028*** 0.0229*** 0.5344*** −0.0077 (0.0431) (0.0710) (0.0568) (0.0077) (0.0761) (0.0139) Schooling(ˇ4) 0.0261*** 0.0272*** -0.0086 −0.0835*** 0.0256*** −0.0779*** (0.0077) (0.0094) (0.0134) (0.0046) (0.0092) (0.0076) GDPt−1(ˇ5) −0.0740*** −0.0822*** −0.1204*** 0.0773*** −0.1228*** 0.0150 (0.0125) (0.0155) (0.0220) (0.0192) (0.0194) (0.0264) DCrisis(ˇ6) −0.0804*** −0.0793*** −0.0942*** 0.0423*** −0.0903*** 0.0345*** (0.0050) (0.0061) (0.0088) (0.0063) (0.0062) (0.0092) DLElection(ˇ7) 0.0424*** 0.0427*** 0.0375*** 0.0048* 0.0463*** 0.0024 (0.0063) (0.0082) (0.0090) (0.0027) (0.0083) (0.0029) Intercept(ˇ0) 0.7950*** 0.6586*** 1.5897*** 0.9421*** 1.2110*** 1.4246*** (0.1631) (0.1801) (0.2929) (0.1871) (0.2461) (0.3032) Interaction(ˇ1+ˇ8) 0.0097* 0.0011 0.0422*** −0.0753** 0.0434*** −0.1066 F-statistics 3.12 0.03 16.24 5.65 17.20 0.63 p-Value 0.0777 0.8697 0.0001 0.0176 0.0000 0.4294 AdjustedR2 _0.3315 _0.3703 _0.3396 _0.3379 _0.3841 _0.3229 N 1225 747 478 1225 747 478

Note:Thistablepresentstheestimationsoftheﬁxed-effectsmodelwithacrisisinteractionvariable.Themodelswithstate-ownedbankcreditareestimatedwithOLS.

Becauseoftheendogenityproblem,anIVmethodisusedintheestimationsofmodelswithprivatebankcredit.TheF-statistictestswhether(ˇ1+ˇ8)=0.

alowershareofthecreditmarket.Incrisisyears,thecoefficient onpercapitaprivatebankcreditchangessignificantlyinallthree groupsof provinces.Thecoefficientincreasesin less-developed provinces,whereasitdeclinesinallprovincesandindeveloped provinces.Thisnegativeorinsignificantcoefficientcanbeobserved ifprivatebanksissuecreditincrisisperiodstobeusedasworking capitalratherthanforinvestmentpurposes.

5.4. Bankcreditandlocalgrowthinelectionyears

Table6 presents theresultsof thefixed-effects model with alocalelectionand bankcredit interactionvariable.Theresults areconsistentwiththe previousfindingsthat percapita state-ownedbankcreditispositivelyandsignificantlyassociatedwith localgrowthin allprovincesaswellasin developedprovinces. Eventhoughstate-ownedbankshaveasignificantlyhighershare in credit markets in local election years (Table 3), the regres-sionfindingsindicatethat,controllingforotherfactorsthatmight affectgrowth,thecoefficientofpercapitacreditprovidedby state-ownedbanksdoesnotchangesignificantlyintheseyears.When thecombinedeffectsareconsidered,theresultssuggestthatper capitacreditfromstate-ownedbanksinlocalelectionyearsdoes notbenefitindividuals’well-beinginanyTurkishprovinces.This findingsupports theotherfindings in theliterature that state-ownedbanksmight providecredit forpolitical reasonsin local elections.

To further investigate the insignificant coefficient of state-ownedbankcreditpercapitaduringelectionperiods,werunour modelforasampleofprovinceswheretheincumbentisamember ofthegoverningpartyintheyearofalocalelection.Inthisway,we aimtoexaminewhetherfundingbystate-ownedbanksismainly usedtoprovidebenefitstoprovincesthatarepopulatedwith sup-portersofthegoverningparty.Wefindthatthecoefficientonper capitastate-ownedbankcreditispositiveandsignificant,butin

localelectionyearsthiscoefﬁcientbecomessigniﬁcantand neg-ative,whichsuggeststhatstate-ownedbanksprovidepolitically connectedcreditduringlocalelectionyears.23

Whenweexaminepercapitaprivatebankcredit,its associa-tionwithlocalgrowthisfoundtobepositiveandsignificantinall provincesandinless-developedprovinces.Thiscoefficientdeclines significantlyinlocalelectionyears.However,thecombinedeffect ofcreditintheseyearsisfoundtobeinsignificantinallprovinces regardlessoftheirdevelopmentlevel.

Table7presentstheresultsofthefixedeffectsmodelwitha generalelectioninteractionvariable.Consistentwithother mod-els,state-ownedbankcreditpercapitaisfoundtoincreaselocal growthratesignificantlyinallanddevelopedprovinceswhereas privatebankcreditper capitais foundtoincreaselocalgrowth significantlyinless-developedprovinces.Althoughthecoefficient onthe interactionvariable is foundtobenegative and signifi-cantinsomeprovinces,theonlysignificantandpositivecombined coefficientisobserved forprivatebankcredit inless-developed provinces.Ingeneral,creditfrombothtypesofbanksisnot signifi-cantlyassociatedwiththegrowthrateofpercapitarealGDPinthe Turkishprovincesingeneralelectionyears.Thisinsignificant find-ingmightbeexplainedbythefirms’behavior.Theymightpostpone theirinvestmentsduringelectionperiods.JulioandYook(2012) examine248electionsin48countriesandfindthatfirmsreduce investmentexpendituresduringnationalelectionyearsbyan aver-ageof 4.8%,controllingfor growthopportunitiesand economic conditions. Theyargue that political uncertainty creates uncer-taintyaboutfutureinvestmentpayoffs;and,inresponse,rational managerspostponeinvestmentsuntiltheuncertaintyisresolved. OurfindingssuggestthatTurkishfirmsmightbehaveinasimilar

23_We_do_not_report_the_results_of_this_estimation_to_save_space._They_are_available

(10)

Table6

Bankcreditandlocalgrowthinlocalelectionyears.

BankCredit(ˇ1) 0.0090** 0.0047 0.0126** 0.0081* 0.0224*** 0.0027

(0.0038) (0.0052) (0.0054) (0.0044) (0.0075) (0.0051)

DLElection×BankCredit(ˇ8) −0.0016 0.0017 −0.0043 −0.0824*** 0.0014 −0.1100***

(0.0104) (0.0112) (0.0411) (0.0138) (0.0363) (0.0223) PublicExpenditures(ˇ2) 0.0643*** 0.0922*** 0.0101 0.3629*** 0.1087*** 0.3485*** (0.0188) (0.0272) (0.0250) (0.0504) (0.0268) (0.0693) Urbanization(ˇ3) 0.3396*** 0.5865*** 0.3044*** 0.0227*** 0.5034*** −0.0079 (0.0431) (0.0705) (0.0575) (0.0077) (0.0758) (0.0139) Schooling(ˇ4) 0.0261*** 0.0273*** −0.0068 −0.0793*** 0.0258*** −0.0738*** (0.0077) (0.0094) (0.0136) (0.0040) (0.0093) (0.0058) GDPt−1(ˇ5) −0.0741*** −0.0818*** −0.1172*** 0.0784*** −0.1147*** 0.0153 (0.0125) (0.0155) (0.0222) (0.0192) (0.0193) (0.0266) DCrisis(ˇ6) −0.0799*** −0.0823*** −0.0723*** 0.0453*** −0.0802*** 0.0354*** (0.0039) (0.0052) (0.0058) (0.0072) (0.0052) (0.0117) DLElection(ˇ7) 0.0431*** 0.0419*** 0.0397* −0.0040 0.0470*** −0.0006 (0.0077) (0.0093) (0.0220) (0.0049) (0.0123) (0.0050) Intercept(ˇ0) 0.7965*** 0.6484*** 1.5355*** 0.9493*** 1.1222*** 1.4449*** (0.1632) (0.1803) (0.2960) (0.1873) (0.2457) (0.3026) Interaction(ˇ1+ˇ8) 0.0074 0.0064 0.0083 −0.0743 0.0238 −0.1073 F-statistics 0.49 0.31 0.04 0.41 0.41 0.10 p-Value 0.4859 0.5771 0.8413 0.5226 0.5221 0.7508 AdjustedR2 _0.3315 _0.3695 _0.3235 _0.3365 _0.3766 _0.3218 N 1225 747 478 1225 747 478

Notes:Thistablepresentstheestimationsoftheﬁxed-effectsmodelwithalocalelectioninteractionvariable.Themodelswithstate-ownedbankcreditareestimatedwith

OLS.Becauseoftheendogenityproblem,anIVmethodisusedintheestimationsofmodelswithprivatebankcredit.F-statistictestswhether(ˇ1+ˇ8)=0.

wayduringgeneralelectionperiods.AsseeninTable2,bothtypes ofbankslendlessingeneralelectionyears.

Itcanbearguedthattheremaybeanendogeneityproblemwith thegeneralelectionvariableiftheelectionwasconductedoutside

theusualschedulebecauseofeconomicturmoil.Duringour sam-pleperiod,allgeneralelectionsinTurkeywereconductedintheir scheduledyear(i.e.,everyfouryears),exceptfortheplanned2002 election,whichwasconductedoneyearearliertoresolvepolitical

Table7

Bankcreditandlocalgrowthingeneralelectionyears.

All Less-developed Developed All Less-developed Developed

BankCredit(ˇ1) 0.0082** 0.0044 0.0142** 0.0035 0.0173** −0.0002

(0.0040) (0.0054) (0.0058) (0.0044) (0.0078) (0.0051)

DGElection×BankCredit(ˇ8) −0.0054 −0.0007 −0.0214** −0.0773*** 0.0040 −0.1022***

(0.0060) (0.0078) (0.0093) (0.0141) (0.0090) (0.0228) PublicExpenditures(ˇ2) 0.0475** 0.0707** −0.0074 0.4128*** 0.0822*** 0.3741*** (0.0194) (0.0279) (0.0257) (0.0508) (0.0274) (0.0699) Urbanization(ˇ3) 0.3613*** 0.6304*** 0.2944*** 0.0269*** 0.5704*** −0.0064 (0.0438) (0.0709) (0.0590) (0.0079) (0.0759) (0.0142) Schooling(ˇ4) 0.0291*** 0.0316*** −0.0105 −0.0862*** 0.0318*** −0.0785*** (0.0079) (0.0096) (0.0141) (0.0041) (0.0097) (0.0060) GDPt−1(ˇ5) −0.0733*** −0.0851*** −0.1049*** 0.0565*** −0.1128*** −0.0005 (0.0128) (0.0157) (0.0228) (0.0195) (0.0199) (0.0268) DCrisis(ˇ6) −0.0863*** −0.0897*** −0.0773*** −0.0090* −0.0889*** 0.0039 (0.0041) (0.0054) (0.0060) (0.0050) (0.0054) (0.0077) DGElection(ˇ7) −0.0070 −0.0138** 0.0098 −0.0009 −0.0184*** −0.0045 (0.0051) (0.0064) (0.0081) (0.0030) (0.0070) (0.0032) Intercept(ˇ0) 0.7709*** 0.6593*** 1.3853*** 0.8553*** 1.0481*** 1.3271*** (0.1668) (0.1830) (0.3030) (0.1915) (0.2514) (0.3091) Interaction(ˇ1+ˇ8) 0.0028 0.0037 −0.0071 −0.0738 0.0213** −0.1024 F-statistics 0.20 0.20 0.62 0.26 3.89 0.65 p-Value 0.6568 0.6580 0.4329 0.6081 0.0490 0.4221 AdjustedR2 _0.3081 _0.3513 _0.3060 _0.3133 _0.3559 _0.3036 N 1225 747 478 1225 747 478

Note:Thistablepresentstheestimationoftheﬁxed-effectsmodelwithageneralelectioninteractionvariable.Themodelswithstate-ownedbankcreditareestimatedwith

OLS.Becauseoftheendogenityproblem,anIVmethodisusedintheestimationsofmodelswithprivatebankcredit.TheF-statistictestswhether(ˇ1+ˇ8)=0.

(11)

Table8

Robustnesschecks.

All Less-developed Developed All Less-developed Developed

PanelA:BankcreditisdeﬁnedasapercentageofprovincialGDP

A.1BankCredit/GDP 5.038 −28.035 105.875 47.808 189.579** _48.851

(28.428) (31.768) (69.765) (37.143) (87.176) (38.981)

A.2BankCredit/GDP 44.667 10.166 58.282 25.546 151.896* _34.552

(37.757) (44.686) (72.058) (39.615) (88.508) (42.078)

DCrisis× BankCredit/GDP −67.100 −57.045 365.363** 58.835 297.472** 34.411

(42.111) (46.946) (149.531) (36.582) (131.808) (38.081)

Interaction −22.433 −46.879 423.645*** _84.381* _449.368*** _68.964

A.3BankCredit/GDP 4.617 −28.307 104.175 46.750 196.456 48.650

(28.452) (31.799) (70.085) (37.195) (89.279) (39.171)

DLElection×BankCredit/GDP 53.829 39.916 124.315 −55.697 −93.489 −5.894

(112.838) (119.631) (430.488) (91.859) (258.508) (99.175)

Interaction 58.446 11.608 228.489 −8.947 102.967 42.756

A.4BankCredit/GDP 1.290 −34.424 136.100* _30.171 _161.728* _44.301

(30.706) (34.099) (76.991) (38.115) (90.916) (39.769)

DGElection×BankCredit/GDP −22.281 4.326 −240.366* -57.511 −1.516 −106.670**

(49.844) (54.661) (128.529) (43.686) (131.343) (45.428)

Interaction −20.992 −30.099 −104.266 -27.339 160.212 −62.369

PanelB:Thelocalelectiondummyvariabletakesavalueof1forelectionyearsandtheyearbeforetheelectionand0otherwise

B.1BankCredit 0.004 0.002 0.005 −0.002 0.016* _0.001

(0.004) (0.006) (0.006) (0.003) (0.009) (0.003)

DNewElection×BankCredit 0.015** 0.011 0.019 0.003 0.055** 0.004

(0.006) (0.008) (0.012) (0.004) (0.023) (0.004)

Interaction 0.018*** _0.013 _0.024* _0.002 _0.071*** _0.005

PanelC:ProvincialGDPgrowthrateisthedependentvariable

C.1BankCredit 0.013*** _0.013*** _0.009* _0.002 _0.020*** _0.002

(0.003) (0.004) (0.005) (0.002) (0.007) (0.002)

C.2BankCredit 0.012*** _0.013*** _0.006 _0.002 _0.021*** _0.002

(0.004) (0.005) (0.006) (0.002) (0.007) (0.002)

DCrisis×BankCredit 0.002 −0.002 0.029*** −0.002 −0.005 0.002

(0.005) (0.006) (0.010) (0.002) (0.008) (0.003)

Interaction 0.015*** _0.012** _0.035*** _0.000 _0.015 _0.003

C.3BankCredit 0.014*** _0.013*** _0.011** _0.001 _0.020*** _0.002

(0.003) (0.005) (0.005) (0.002) (0.007) (0.002)

DLElection×BankCredit −0.004 −0.003 −0.001 −0.003 0.007 0.000

(0.009) (0.010) (0.039) (0.004) (0.029) (0.005)

Interaction 0.010 0.010 0.010 −0.001 0.027 0.002

C.4BankCredit 0.014*** _0.014*** _0.013** _0.000 _0.018** _0.001

(0.004) (0.005) (0.006) (0.002) (0.007) (0.002)

DGElection×BankCredit −0.008 −0.006 −0.020** 0.003 0.006 −0.003

(0.005) (0.007) (0.009) (0.002) (0.008) (0.003)

Interaction 0.006 0.008 −0.007 0.003 0.025** _−0.002

PanelD:Parameterestimateswithyeardummyvariables

D.1BankCredit 0.011*** _0.008 _0.017*** _0.027*** _0.051*** _0.016***

(0.004) (0.005) (0.006) (0.004) (0.006) (0.004)

D.2BankCredit 0.011*** _0.011** _0.013** _0.026*** _0.047*** _0.015***

(0.004) (0.006) (0.006) (0.004) (0.007) (0.005)

DCrisis×BankCredit 0.000 −0.008 0.032*** 0.003 0.010 0.002

(0.005 (0.007) (0.010) (0.003) (0.009) (0.003)

Interaction 0.011** _0.003 _0.045*** _0.029*** _0.057*** _0.017***

D.3BankCredit 0.011*** _0.008 _0.017*** _0.028*** _0.051*** _0.016***

(0.004) (0.005) (0.006) (0.004) (0.006) (0.004)

DLElection×BankCredit −0.003 0.000 −0.003 −0.005 0.007 −0.001

(0.010) (0.011) (0.039) (0.005) (0.033) (0.005)

Interaction 0.008 0.008 0.014 0.023*** _0.058* _0.015**

D.4BankCredit 0.012*** _0.009* _0.018*** _0.027*** _0.047*** _0.016***

(0.004) (0.005) (0.006) (0.004) (0.006) (0.004)

DGElection×BankCredit −0.005 −0.007 −0.007 0.001 0.069*** −0.002

(0.006) (0.008) (0.010) (0.003) (0.017) (0.003)

Interaction 0.007 0.002 0.011 0.028*** _0.116*** _0.014*

Note:Thenumbersinparenthesesrepresentstandarderrors.Interactionrepresentsthecoefﬁcientonthecreditvariableincrisisorelectionyears.

(12)

instability. Whenwe estimate ourmodel excluding the obser-vationsin2002,thesignificanceofourresultsdoesnotchange. Thepositiveandsignificantassociationbetweenlocalgrowthrate andstate-ownedbankcreditisobservedinallprovincesandin developedprovinces,andlikeotherfindings,theirsignificance dis-appearsingeneralelectionyearsbutremainsincrisisyears.

Thefindingswithrespecttocontrolvariablesaresimilartothe otherstudiesintheliterature.Interestingly,thegrowthrateofGDP percapitaissignificantlyloweringeneralelectionyearsin less-developedprovinces,althoughtheygenerallyhaveasignificantly highergrowthrateinlocalelectionyears.

5.5. Robustnesschecks

Asarobustnesscheck,weredefinebankcreditasa percent-ageofGDPratherthanpercapita.PanelAofTable8reportsonly thecoefficientsofbankcreditvariables.Nosignificantcoefficient onstate-ownedbankcredit isobservedfor allprovinces. How-ever,creditprovidedbystate-ownedbanksduringcrisisyearsis foundtoincreaselocalgrowthindeveloped provinces. Further-more,althoughstate-ownedbankcredithasapositiveeffecton growthintheTurkishprovinces,theeffectdeclinessignificantly ingeneralelectionyears.Ontheotherhand,privatebankcredit hasapositiveandsignificantassociationwiththegrowthrateof less-developedprovinces.Thepositiveandsignificantcoefficient onprivatecreditisobservedincrisisperiodsaswell.

Itcanbearguedthatiflocalelectionsoccurintheﬁrsthalfof theyear,thegovernmentmightstartelection-relatedspendingin theyearpriortoelection.Therefore,weredeﬁnethelocalelection dummyvariablebyincludingtheyearbeforethelocalelections andestimateourmodelswiththisnewvariable,DNewElection.Panel

BofTable8presentstheestimatedcoefficientsoncredit-related variablesandthetotaleffectofcreditduringlocalelectionyears. Wefindthatcreditprovidedbystate-ownedbanksincreaseslocal growthrateinlocalelectionyearsinallprovincesanddeveloped provinces.However,noneofthestate-ownedbankcredit-related variablesisfoundtosignificantlyaffectgrowthrateinthe less-developedprovinces.

Ontheotherhand,creditissuedbyprivatebanksinpre-election andlocalelectionyearsispositivelyandsignificantlyassociated withpercapitaGDPgrowthrateofless-developedprovinces. How-ever,nosignificantcoefficientofpercapitacreditissuedbyprivate banksis foundinalland developed provinces.Thisresult sup-portsourpreviousfindingsandtheexistenceofapoliticalbusiness cycle.Governments seemtoengageinexpansionaryfiscal poli-ciesbeforeelectionstoenhancetheirre-electionprospects,and pursuecontractionarypoliciesafterelections.InTurkey,alllocal electionsareinthefirstquarteroftheyear,hence,fiscal contrac-tions(Schuknecht,1996)orincreasinguncertainty(JulioandYook, 2012)mightlowerinvestmentsandbankborrowinginelection years.

Todeterminewhetherinternalmigrationamongregionsaffects ourresults,weestimateourmodelusingtherealGDPgrowthrate attheprovinciallevelasanotherdependentvariable.Althoughthe qualityanddirectionofmigrationvarybyperiodinTurkey,the movementfollowsapatternfromless-developedregionstomore developedregions.Becauseoftheunavailabilityofyearly inter-nalmigrationdatabefore2008,wecouldnotcontrolformigration directly;instead,weincludethegrowthrateinpopulationin addi-tiontotheothercontrolvariablesinthemodel.Thecoefficients oncreditvariables andinteractioneffectsarereportedinPanel C ofTable8.A positiveandsignificantcoefficientonpercapita state-ownedbankcreditisobtainedforallprovinces,regardlessof theirdevelopmentlevel.Thetotaleffectofstate-ownedbankcredit incrisisyearsisfoundtobepositiveandsignificantindeveloped

provinces,supportingourpreviousfindingsofpercapitagrowth rate.Ontheotherhand,creditprovidedbystate-ownedbanks dur-inggeneralelectionyearsisnegativelyandsignificantlyrelatedto thegrowthrateofincomeindevelopedprovinces.Interestingly, thecoefficientonprivatebankcreditisfoundtobepositiveand significantonlyinless-developedprovinces.

Wealsoestimateourmodelsbyincludingyeardummyvariables randomlytothemodel(PanelDinTable8).Ourresultswithrespect tostate-ownedbankcreditarefoundtoberobusttothe inclu-sionofyeareffects.Forexample,thecoefficientonstate-owned bankcreditispositiveandsignificantinallprovincesandin devel-opedprovinces;aswellasduringcrisisperiodsintheseprovinces. However,allthecoefficientsonprivatebankcreditarefoundtobe significantandpositiveinallmodels,includingcrisis,localelection, andgeneralelectionperiods.

6. Conclusion

Weexaminepercapitacreditissuedbystate-ownedandprivate banksinTurkeytounderstandtheirroleinthegrowthofTurkish provinceswithdifferentdevelopmentlevels,especiallyincrisisand electionyears.Similartomostotherdevelopingcountries, state-ownedbanksconductasignificantamountofthebankingbusiness inTurkey.Althoughthesebanksoperateinsimilarfieldsofbusiness asprivatebanks,theremainingdifferencesareconsiderable(see, forexample,Hanson,2004).Forexample,lowrefinancingcostsand implicitgovernmentguaranteesallowthemtoissueriskierloans thanprivatebankscan.Statebanksmightprovidefinancing on non-commercialtermstoeconomicallyunderdevelopedregions. Theymightfill thecreditgapduringdistressperiods.However, thecurrent empirical literature shows that state-owned banks aroundtheworldarecharacterizedbypoliticalinterference,low accountabilitytostakeholders,inadequateprudentialregulation andsupervision,andalackofproperincentives.

Inthispaper,wefindthatstate-ownedbanksinTurkeyhavea significantlyhighershareinthecreditmarketsofless-developed regionsinlocalelectionyearsandcrisisperiods.Theresultsofthe fixed-effectsmodelsindicatethatpercapitaprovincialcredit pro-videdbycommercialbanksregardlessoftheirownershipisfound tohaveapositiveandsignificanteffectonlocaleconomicgrowth. Althoughpercapitacreditprovidedbystate-ownedbanksincrisis yearshasasignificantandpositiveassociationwithlocalgrowth inallanddevelopedprovinces,percapitacreditprovidedby pri-vatebanksispositivelyassociatedwiththegrowthrateoflocalper capitaincomeincrisisyearsandinless-developedprovinces. Over-all,ourresultssuggestthatstate-ownedbanksmightissueloansfor politicalreasonsduringelectionperiods,buttheystillcontinueto playasignificantroleinoffsettingtheadverseeffectsofeconomic shocks,especiallyinthedevelopedregions.Althoughacrisisisa rareevent,thesebanksseemtostillbeimportantforlocalgrowth. Theinsignificantassociationbetweenprovincialgrowthrate and per capita credit issued by private banks in developed provincesandthatbetweenprovincialgrowthandpercapitacredit issuedbystate-ownedbanksinless-developedprovincescanbe explainedbytheinefficiencyof thesebanksin the correspond-ingprovinces.Hasanetal.(2009)statethatbanksmightpromote productivitygrowththroughthreeways:morecredit,more effi-cientbanking,andtheirinteraction.EventhoughTurkish banks providemorecreditintheseprovinces,theymaynotbeefficient. Asafurtherstudy,ifbankefficiencymeasuresataprovinciallevel areavailableinthefuture,theimpactofbankefficiencyandits interactionwithcreditcanbeexaminedtoexplainwhyweobserve insignificantassociationinprovinceswheremorecredithasbeen providedbystate-ownedbanks.