Upload
others
View
13
Download
0
Embed Size (px)
Page | 1
Corruption,PoliticalInstitutions,andAccountingEnvironment:ACross‐countryStudy
MuhammadNurulHouqeSchoolofAccounting&CommercialLaw
VictoriaUniversityofWellington,NewZealandPhone:+6444636591,Fax+6444635966
RezaMonem*GriffithBusinessSchoolGriffithUniversity
Brisbane,Queensland,AustraliaTel+61(0)737353598,Fax:+61(0)737353719
Email:[email protected]
Paperpresentedatthe2013annualTheInternationalJournalofAccountingSymposium,17–20May,Wuhan,P.R.China
*Correspondingauthor.
Page | 2
Corruption,PoliticalInstitutions,andAccountingEnvironment:ACross‐countryStudy
Abstract:
Using data from 166 countries over the period 1996‐2011, we investigate the role of
accountinginformationinreducingcorruptionaftercontrollingfortheeffectsofpolitical
institutions and economic development. We find strong evidence that accounting
environmentplaysonlyaminorrolerelativetothatofthestrengthofpoliticalinstitutions
inthecontrolofcorruption.Ourresultsholdevenaftercontrollingforvariablesrelatedto
investorprotectionandculturaldimensions.Ourresultschallengetheviewthatcountries
intendingtoreducecorruptionshouldinvestinhigher‐qualityaccountingstandards. Our
findingsalsosuggestthatcountrieswiththestrongestpoliticalinstitutionsstandtobenefit
mostfromIFRSadoption.
JELclassifications:M41,G38,K42
Keywords: Corruption; Accounting environment; Political institutions; IFRS; Economic
development;Investorprotection
Page | 3
Corruption,PoliticalInstitutions,andAccountingEnvironment:ACross‐countryStudy
1.Introduction
Corruption is a fundamental threat to the viability of any economic system (e.g.,
Blackburn,Bose,&Haque,2008;Mo,2001;Shleifer&Vishny,1993;Wei,2000).Although
avastliteratureexistsonthecausesandconsequencesofcorruptionandhowtocontrolit
(e.g.,Ades&DiTella,1996;Fisman&Svensson,2007;Rock&Bonnett,2004;Tanzi,1998;
Treisman, 2007), research literature linking corruption with accounting is sparse.
However,inarecentstudy,Malagueño,Albrecht,Ainge,andStephens(2010)findthatthe
Big4marketshareandperceivedaccountingqualityareinverselyrelatedtotheperception
ofcorruptioninacountry.Hence,Malagueñoetal.(2010,p.388)conclude,“Countriesmay
beabletodecreasecorruptionbyimprovingthequalityoftheiraccountingandauditing.”
Such a conclusion is premature and misleading given that the weakness of political
institutions is a fundamental andprobably themost crucial element in the economicsof
corruption(Aidt,2003;Lederman,Laoyza,&Soares,2005;Tanzi,1998).Malagueñoetal.
donot fullycontrol for theeffectivenessofpolitical institutions inacountryandmostof
theirproxiesforpoliticalinstitutionsareeitherinsignificantorweaklysignificant.1
In thisstudy,we investigatetheroleofaccounting information inreducingcorruption
aftercontrollingforpoliticalandeconomicfactorsassociatedwithcorruption.Specifically,
wearguethat,althoughtheaccountingenvironmentofacountrymayhavesomeabilityto
1Severalof theproxies forpolitical institutionsused inMalagueñoetal. (2010)arecommon lawsystem,formerBritishcolony,federalsystem,ethnolinguisticdivision,fuelmetalandmineralexports,uninterrupteddemocracy, government intervention, and government turnover. Further, their measure of economicfreedom, following DiRienzo, Das, Cort, and Burbridge (2007), is a mixture of economic and politicalindicators.
Page | 4
reduce perceived corruption, strengthening of the political institutions has the greatest
potential forreducingcorruption. Given thataccountingquality isdirectly influencedby
thestrengthofpoliticalinstitutions(butnotviceversa),countriesaroundtheworldwould
bemuch better off by improving the quality of political institutions in the fight against
corruption.Furthermore,tryingtoimproveaccountingandauditingenvironmentwithout
any concurrent changes in political institutionsmay have theminimal effect in the fight
againstcorruption.
We exploit country‐level data related to control of corruption (corruption perception
index)andpolitical institutions compiledbyKaufmann,Kraay, andMastruzzi (2012). In
thisstudy,proxiesforpoliticalinstitutionsarevoiceandaccountability,andruleoflaw.We
attempt to capture the accounting environment of a country by identifyingwhether the
countryhasadoptedtheInternationalFinancialReportingstandards(IFRS),sourcedfrom
DeloitteIASPluswebsite(2012),andbyutilizingthefinancialdisclosureindexcompiledby
theWorldBank (2012). Our analysis based on data from166 countries over the period
1996‐2011suggeststhataccountingenvironmentplaysaminorrolesecondarytopolitical
institutions in the control of corruption. Our results hold even after controlling for
variables related to economic development, investor protection, and Hofstede’s (2001)
cultural dimensions. Further, our results are robust to alternative specifications of our
models,alternativesamplespecification,andalternativeestimationtechniques.
Our results imply that the countrieswhich intend to reduce corruptionwill bemuch
better off by investing in the improvement of political institutions. Moreover, countries
with weak political institutions that have adopted the IFRS cannot expect to achieve a
Page | 5
reduction in corruption (via improved financial reporting) until political institutions are
strengthened.
We make three contributions to the literature on corruption and international
accounting. First,we contribute to the cross‐country studies on corruptionby specifying
theroleofaccountinginthecontrolofcorruption.Specifically,unlikeKimbro(2002)and
DiRienzo et al. (2007), we control for endogenous relation among political institutions,
corruption,andaccountingenvironment.Further,unlikeMalagueñoetal.(2010),wefind
that political institutions have the strongest effect in the control of corruption. Second,
becauseouranalysisencompasses theperiodwhentheIFRShavebeenadopted inmany
countries,we contribute to the literatureon thebenefitsof IFRSadoption. Our analysis
suggests that, countries with weak political institutions (including weak investor
protection)havelittletogainfromtheIFRSadoption.Third,ourresultsmayhelpexplain
themixed evidence in cross‐country studies that address the effect of IFRS adoption on
financial reporting quality (e.g., Ahmed, Neel, &Wang, 2013; Barth, Landsman, & Lang,
2008;Jeanjean&Stolowy,2008;Soderstrom&Sun,2007). Unlessdifferencesinpolitical
institutions across countries are adequately controlled for, evidence from cross‐country
studiesonfinancialreportingqualityandIFRSadoptionwillremaininconclusive.
The rest of the paper is organized as follows. Section 2 provides an overview of the
nature of corruption. Section 3 briefly reviews some key studies on corruption and
develops the theoretical framework linking corruption and accounting environment.
Section4proposestheresearchmodels,andexplainsthedataandthevariablesusedinthe
Page | 6
study.Empiricalresultsarepresentedinsection5.Section6discussesseveralrobustness
checks.Section7summarizesthepaperandprovidessomeconclusions.
2.Thenatureofcorruption
FollowingShleiferandVishny(1993),wedefinecorruptionbroadlyastheuseofpublic
officeforunauthorizedprivategain.Ourdefinitionofcorruptionisconsistentwiththoseof
Blackburnetal.(2006),andEverett,Neu,andRahaman(2007):corruptionistheabuseof
authority by bureaucratic officials who exploit their discretionary power, delegated to
thembythegovernment,toadvancetheirowninterestsbyengaginginunauthorizedrent‐
seekingactivities.
Aidt (2003)and Jane(2001)argue thatcorruptionhas threeelements.First, someone
must have discretionary power or authority to design regulation or administer policy
outcome. Second, there must be economic rents associated with this power. Third, it
requires a legal or judicial system that decreases the probability to detect and punish
corruptofficials.Hence,opportunities forcorruptionarisewhenever theofficials’actions
involve the exercise of discretion and are impossible to be monitored perfectly (Rose‐
Ackerman,2003).
In the market for corruption, both sides of a transaction have to agree to corrupt
practices(Jane,2001).Forexample,themanagersofprivatefirms,throughtheircontrolof
themanagementandfirmresources,maybribepublicofficialsforsecuringpublicprojects.
Ontheotherhand,publicofficials,actingonself‐interest,mayacceptbribesorkickbacks
from citizens and corporations to maximize their own wealth (Rose‐Ackerman, 2003).
Further, inorder for thecorruption to takeplace,bothpartiesmustagree toconceal the
transactionbecauseitsrevelationcanhavepunitiveconsequencesforbothparties.
Page | 7
Corruption has devastating effects, especially on the citizens of developing countries.
Corruptionreduceseconomicgrowth(Mauro,1995;Mo,2001;Habib&Zurawicki,2002;
Zhao, Kim, & Du, 2003). An estimate of the effect of corruption on economic growth is
providedbyMo(2001):a1%increaseinthecorruptionlevelreducesthegrowthrateby
about0.72%.Highlycorruptgovernmentsalsospendlessoneducationandhealth(Tanzi,
1998), thereby limiting the potential for economic growth. Mauro (1997) claims that
corruption may reduce the efficiency of domestic and international aid flow through
diversion of funds from intended government projects. While Kehoe (1998) notes that
corruptpracticeselevatethehiddencostofdoinginternationalbusiness,GhosalandMoran
(2005) indicate thatmultinationalcorporationssuffer tarnishedreputations in theworld
marketplace when they forgo their social legitimacy by engaging in corrupt practices.
Furthermore,corruptionreducesrevenuegeneratedthroughtaxationwhenpartiesengage
in tax evasion, contributing to adverse budgetary consequences for the government
(Mauro,1997).
3.Priorstudiesandtheoreticalframework
3.1.Priorstudies:Causesandconsequencesofcorruption
The literature on corruption is enormous and still growing. Besides, several excellent
reviewpapersareavailableonthecausesandconsequencesofcorruption(e.g.,Treisman,
2007;Lambsdorf,2005).Hence,wearegoingtohighlightonlysomeofthekeyfindingsin
thisarea.
SandholtzandKoetzle(2000)arguethat thepolitical‐economicstructureof incentives
andopportunities,andthepeople’sculturalorientationsarethetwoprimaryfactorsthat
determinethelevelofcorruptioninacountry. TheirargumentisreflectedinTreisman’s
Page | 8
(2007)observations.Basedon a surveyof a decadeof cross‐national empirical studies
since the mid‐1990s, Treisman (2007, p. 211) observes, “[H]ighly developed, long‐
establishedliberaldemocracies,withafreeandwidelyreadpress,ahighshareofwomen
ingovernment,andahistoryofopennesstotrade,areperceivedaslesscorrupt.”Further,
Lederman et al. (2005) document that political institutions in the form of democracy,
parliamentary system, political stability, and freedomof press are associatedwith lower
corruption.TheseviewsareconsistentwiththefindingsofPaldam(2002),andAliandIsse
(2003). In particular, Ali and Isse (2003) find that education, judicial efficiency, and
economic freedom are negatively related to corruption while a country’s foreign aid
dependencyandthesizeofgovernmentarepositivelyrelatedtocorruption.Insum,there
isoverwhelmingevidencetosuggestthatthestrengthofpoliticalinstitutionsinacountry
reallymatters in thecontrolof corruption.Hence, agovernmentwhichwishes to reduce
corruption“shouldsettleforsimpleandstablelegalandadministrativerulesandimprove
ontheinformationprovidedtotheprivatesector”(Lambert‐Mogiliansky,2002,p.48).
In terms of economic determinants of corruption, the most significant (economic)
determinant of corruption is the real gross domestic product (GDP) per capita (Paldam,
2002). Further, within a general equilibrium context, Blackburn et al. (2006, 2010)
demonstrate that corruption and economic development are endogenously determined
with a negative relationship between them. In one of the early studies, Tanzi (1998)
identifieseconomicgrowthasoneof themajorcostsofcorruptionandarguesthatsome
forms of state reforms are required to lower the supply of and demand for corruption.
Further,themostprominentchannelthroughwhichcorruptionaffectseconomicgrowthis
politicalinstability(Mo,2001).Othernegativeeffectsofcorruptionincludeslowingdown
Page | 9
of foreign direct investment (FDI) in the host country (Shleifer and Vishny, 1993;
Smarzynska & Wei, 2000; Wei, 2000) and misallocation of resources in an economy
because of the necessary secrecy of corruption (Ehrlich & Lui, 1999; Shleifer & Vishny,
1993). Further, corruption in thehost country shifts theownership structure related to
FDItowardsjointventures(Smarzynska&Wei,2000).Evidencealsoexiststhatcorruption
inverselyaffectstherealexchangerateofacountry(Bahmani‐Oskooee&Nasir,2002).
While corruption slows economic growth and investments in most countries, the
economic growth of the large East Asian countries despite high‐level corruption is a
paradox(Rock&Bonnett,2004).However,theEastAsianparadoxcanbeexplainedbythe
highpredictabilityofcorruptionintheregionbecausecorruptionregimeswhicharemore
predictablehavelessnegativeimpactoninvestment(Camposetal.,1999).Priorresearch
hasalsodocumentedfirm‐levelconsequencesofcorruption.Usingdatafromthreeworld‐
wide surveys, Kaufmann andWei (1999) find that firmswhich paymore bribes end up
spendingmoremanagement timewith bureaucrats on negotiating regulations, and face
higher cost of capital. Further, Fisman and Svensson (2007) document that bribery is
negativelyrelatedtofirmgrowth.Insum,corruptionhasmanyadverseconsequencesfor
aneconomy,includinglowereconomicgrowthrate,lowerFDI,misallocationofresources,
and weaker foreign exchange rate. Further, corruption and economic development are
negativelyandendogenouslyrelatedthroughthecompetitionbetweengrowth‐enhancing
andsociallyunproductiveinvestments(EhrlichandLui,1999).
Page | 10
3.2.Priorstudies:Corruptionandaccounting
Asalreadystated,the literature linkingcorruptionandaccounting issparse. Wewere
abletoidentifyonlythreestudiesinthisarea.TheseareKimbro(2002),Malagueñoetal.
(2010), and Wu (2005a). A fourth study (DiRienzo et al., 2007), although not directly
related to accounting, reports that digital access to information can lower corruption.
Becauseoneoftherolesoffinancialreportingistoreduceinformationasymmetrybetween
managers who control the firms and owners who supply the capital, accounting
environmentinacountryislikelytoplayaroleinthecontrolofcorruption.
Both Kimbro (2002) and Malagueño et al. (2010) conclude that the perception of
corruption is negatively related to accounting quality. Kimbro (2002) employed the
corruption perception index of the years 1995 to 1999 published by the Transparency
International.Shemeasuredaccountingqualityasthenumberofaccountantsper100,000
inhabitants and the CIFAR reporting index based on financial statements in 1990.
Malagueño et al. (2010) used the Big 4 market share and perceived accounting quality
(PAQ),sourcedfromtheWorldEconomicForum(2003),asproxiesforaccountingquality.
Finally,usingcross‐country firm‐leveldata inAsiansetting,Wu(2005) found thatbetter
accounting practices can help reduce both the incidence of bribery and the amounts of
bribe payments, but conforming to high quality accounting standards alone does not
necessarily bring down the incidence of bribery. Wu’s findings support the notion that
accounting can play a role in controlling corruptionwhen other institutional factors are
supportive.
Page | 11
Arguably,thenumberofaccountantsper100,000inhabitantsusedbyKimbro(2002)is
a proxy for the level of economic development rather than accounting quality. Further,
Kimbro (2002) used (average) GDP growth rate in only one specification of her main
model (seeTable3,p.339)andthecoefficient(=0.126) is insignificant(t=1.517). Her
otherproxyforeconomicgrowth,D74‐80,isadummyvariablewhichtakesavalueof1if
thegrossnationalproduct(GNP)ofacountry isgreaterthanthesamplemedian. AGNP
greaterthanthesamplemediandoesnotcaptureeconomicgrowth,ratheritcapturesthe
level of economic development. Hence, Kimbro’s conclusion that moderate economic
growthisrelatedtolowerlevelofcorruptionisunwarranted.
3.3.Theoreticalframework:Linkbetweencorruptionandaccountingenvironment
Ifeconomics isaboutexpandingthepieandpolitics isaboutdistributing it (Alesina&
Rodrik,1994),accountingisaboutmeasuringthesizeofthepie.Inthissense,economics,
politics,andaccountingareinter‐related.
ShleiferandVishny(1993)articulatethatmaintainingsecrecyofthe corruption ‘deal’
by both parties involved is a necessary condition for the supply of corruption. Further,
DiRienzoetal.(2007)provideevidencethataccesstoinformationisnegativelyrelatedto
corruption. Astheroleofaccountingis toprovideinformationforefficientallocationof
resourcesinaneconomy,non‐disclosureofsecret‘deals’islikelytoperpetuatecorruption.
On the other hand, high-quality accounting is likely to act as a deterrent on the demand side of
corruption. High-quality accounting information is a product of not only the accounting
standards, but also a host of other factors such as managerial incentives, rule of law, and
enforcement (Ball, Robin and Wu, 2003). Hence, firm managers are less likely to engage in
bribery payment in countries with strong rule of law and enforcement. Further, strong internal
Page | 12
control systems, strong managerial over-sight, and a high degree of accountability are not only
likely to deter bribery payments but also help quick detection of such irregularities. Thus, the
level of perceived corruption and the accounting environment in a country are
endogenously relatedbecause better accounting regime can reduce corruption by better
disclosureofeconomiceventswhilehighercorruptionlevelcanimpedethedevelopment
offinancialreportinganddisclosuretohidecorruption.
4.ResearchDesign
4.1.Data
Fortheprimarydependentvariable,weusecountry‐levelcorruptionperceptionindex
(Low_Corrup) as measured by Kaufmann et al., (2012) in their Worldwide Governance
Indicator (WGI) project. Our initial sample comprises annual observations for 214
countriesovertheyears1996to2011.Fromthisinitialsample,weeliminated31countries
whosecorruptionscoreswerenotavailableinKaufmanetal.(2012).Thenweexcluded17
countries due to missing disclosure index scores, one of our proxies for accounting
environment.Thesetwoexclusionsresultedinasampleof166countries.
Weuseperception‐basedcorruptionindexasameasureofcorruptioninsteadofactual
corruptionexperience.Wemadethischoicebecauseofstrongempiricalevidenceinfavor
of perception‐based measures. Treisman (2007, p. 212) observes, “The more subjective
indexesofperceivedcorruption—basedonevaluationsofexpertsandopinionsofbusiness
people and citizens—turn out to be highly correlated with a variety of factors that are
commonlybelievedtocausecorruption.”Ontheotherhand,measuresofactualcorruption
experience hardly correlate with any of the factors believed to be causes of corruption,
onceonecontrolsforincome(Treisman,2007).
Page | 13
ThevariableLow_Corruprangesfrom‐1.241to2.374,withhigherscoresindicatingless
corruption. Another popular measure of corruption is Transparency International’s
CorruptionPerception Index (TICPI). It is arguablyone of thebest, andwidelyused in
cross‐countryanalysisofcorruption(Malagueñoetal.,2010).However,weuseKaufmann
et al.’s (2012) corruption indexbecause it incorporatesdata frommore sources thanTI
CPIandattempts to improveonthe treatmentofstatisticaluncertainty inTICPI(Knack,
2007).
Wemeasureaccountingenvironment(Acc_Env)asthesumofthescoreswithregardto
whetheracountryadoptedtheIFRS2forexternalreportingbyitsdomesticfirms,andthe
extent of disclosure requirements. We obtain the IFRS adoption data from the Deloitte
IASPlus website (2012), setting the value of one (1) for countries that have adopted the IFRS
and zero (0) otherwise.
(InsertTable1here)
AshbaughandPincus(2001)andDingetal.,(2007)provideevidencethatIFRSrequires
more comprehensive disclosure than most local accounting standards. Other literature
suggeststhatgreaterdisclosureaftertheadoptionofIFRShasasignificantpositiveeffect
on investors’ confidence, by reducing information asymmetry, agency problems, and
reporting uncertainties (e.g., Barth et al., 2008; Hope et al., 2006; Houqe et al., 2012;
Soderstrom&Sun,2007;Wu,2005a;Zarb,2008).Further,Houqeetal.,(2012)arguethat
adopting a common set of accounting standards, such as the IFRS, vigorously forces
managementtoreportfaithfullyandtruthfully.Thus,managerswillengagelessincorrupt
activities.Inthissense,theadoptionofIFRSreflectsahigh‐qualityaccountingenvironment
2 We interpret adoption of IFRS in a broader sense regardless of the process of how such ‘adoption’ took place.
Page | 14
bysettingasystemthatallowsfirmstorecognizeaccountinglossesthatarisefrombribery
orillegalpaymentsinanappropriatemanner.
Unfortunately,thereisremarkablylittleliteraturelinkingtheadoptionofIFRSwiththe
levelofcorruptionincountries.Zarb(2008)documentsthattheuseofIFRSisnegatively
correlated with the perception of corruption in developed countries. Hence, we suggest
that the move towards the global use of IFRS appears to be the new platform for
transparency in financial reporting and is potentially another tool in the fight against
corruption. Further, the transparency in accounting information and disclosure can help
reduce the level of corruption by increasing the probability of corrupt practices being
detected(Wu,2005b).
WecollectedtheextentofdisclosurerequirementsdatafromtheWorldBank(2012).It
measurestheextenttowhichinvestorsareprotectedthroughdisclosureofownershipand
financialinformation.Theindexrangesfrom0to10,withhighervaluesindicatinggreater
disclosure.Bushmanet al., (2004)suggest thathigherdisclosurecanhelp reduceagency
problems between management and shareholders, thus preventing the opportunistic
behaviorofmanagers.
We use voice and accountability, and rule of law as twomeasures of the strength of
political institutions (Pol_Ins) in a country, as per Kaufmann et al. (2012). We use two
controlvariablesinthemainmodels:investorprotectionindex(Inv_Pro)publishedbythe
World Bank (2012), and the level of economic development (Eco_Dev)measured as the
natural logarithm of GDP (in US$) as per the World Bank (2010). Our choice of these
controlvariablesismotivatedbytheextantliterature.LaPorta,Lopez‐de‐Silanes,Shleifer,
&Verdy(1998)documentthatweakinvestorprotectionrightscreateagencycostsinthe
Page | 15
formofexpropriationof shareholder (minority shareholder)wealthby insidermanagers
(majority shareholders). To the extent that, corruption is a form of agency cost, the
corruption level in a country is expected to be inversely related to the level of investor
protection. We include economic development as a control variable because of
overwhelmingevidence,asalreadydiscussedinSection3.1,ofthelinkbetweencorruption
and economic development (e.g., Blackburn et al., 2006, 2010; Paldam, 2002; Treisman,
2007).Table1providesthe listofall thevariables inthisstudy,theirdetaileddefinition,
anddatasources.
4.2.Models
Our objective in this study is to understand the role of accounting environment in
reducing corruption after controlling for variables related to political institutions and
economicdevelopment, rather thanto identifyall thedeterminantsofcorruption.Hence,
weproposeparsimoniousmodels.Ourfirsteconometricmodelisasfollows:
Low_Corrup=α0+α1(Acc_Env)+α2(Inv_Pro)+α3(Eco_Dev)+ε(1)
whereallvariablesareasdefinedearlier.
Inmodel(2)wereplaceAcc_Envwiththestrengthofpoliticalinstitutions(Pol_Ins):
Low_Corrup=ά0+ά1(Pol_Ins)+ά2(Inv_Pro)+ά3(Eco_Dev)+ε(2)
whereallvariablesareasdefinedearlier.
Inmodels(1)and(2),ourintentionistounderstandtheroleofaccountingenvironment
(Acc_Env) and the strength of political institutions (Pol_Ins) separately in the control of
corruption. In model (3), we combine models (1) and (2) to examine the relative
contributionofAcc_EnvandPol_Insinexplainingperceivedcorruption.Thus,model(3)is
asfollows:
Page | 16
Low_Corrup=ψ0+ψ1(Acc_Env)+ψ2(Pol_Ins)+ψ3(Inv_Pro)+ψ4(Eco_Dev)+ε(3)
whereallvariablesareasdefinedearlier.
We estimate models (1) – (3) using the ordinary least squares (OLS) estimation technique.
However, prior research suggests that (DiRienzo et al., 2007; Kimbro, 2002; Malagueño et al.
2010) that Acc_Env,Pol_Ins,andLow_Corrup could all be endogenously related in the sense that
they are jointly determined. Hence, we treat each of these three variables as an endogenous
variable and employ two-stage least squares (2SLS) technique to estimate the models. Our
empirical models are as follows:
Low_Corrup=α0+α1(Acc_Env)+α2(Pol_Ins)+α3(Inv_Pro)+α4(Eco_Dev)+ε(4)
Acc_Env=ψ0+ψ1(Low_Corrup)+ψ2(Pol_Ins)+ψ3(Inv_Pro)+ψ4(Eco_Dev)+ε(5)
Pol_Ins=ά0+ά1(Low_Corrup)+ά2(Inv_Pro)+ά3(Eco_Dev)+ε(6)
whereallthevariablesareasdefinedearlier.
4.3.Descriptivestatistics
Table 2 provides descriptive statistics of the key variables in this study. As Table 2
reveals, the corruptionperception index (Low_Corrup) ranges from ‐1.598 (most corrupt
country) to 2.441 (least corrupt country). Thus, in our sample, the average country is
highlycorruptwithmean(median)scoreof‐0.046(‐0.304)sittingataround38%(32%)
ofthescale.Themean(median)scoreoftheaccountingenvironment(Acc_Env)is6.301
(6.000) in a scale that ranges from 0.000 (weakest) to 11.000 (strongest) accounting
environment.Intermsofthestrengthofpoliticalinstitutions(Pol_Ins),theaveragecountry
liesaround46%(41%)ofthescalewiththemean(median)scoreof‐0.196(‐0.518)ina
scalethatrangesfrom‐3.299(weakest)to3.516(strongest)politicalinstitutions.
(InsertTable2here)
Page | 17
Table3presents abivariatecorrelationmatrixwithPearson’s correlationsbelow the
diagonalandSpearman’scorrelationsabovethediagonal.AsTable3shows,amongallthe
variables,corruptionperceptionindex(Low_Corrup)hasthestrongestpositivecorrelation
with thestrengthofpolitical institutions (r=0.923,p<0.001; ρ=0.911,p<0.001) followed
bythelevelofeconomicdevelopment(r=0.752,p<0.001;ρ=0.686,p<0.001).Amongallthe
variables, accounting environment (Acc_Env) has the weakest, albeit positive and
significant, relation with the corruption perception index (r=0.260, p=0.001; ρ=0.223,
p=0.004). These results suggest that countries which invest in strengthening political
institutionswillachievethegreatestresultsinreducingcorruption.
(InsertTable3here)
5.Results
Table4reportstheOLSestimatesofmodels(1)to(3).Inmodel(1),theadjustedR2 is
60.9%.Further,accountingenvironment (Acc_Env) ispositively related to thecorruption
perceptionindexandthecoefficientissignificantatthe5percentlevel(t‐staticofAcc_Env
=2.044).3Among theothervariables inmodel (1), both investorprotection (t‐statisticof
Inv_Pro=3.834)andthelevelofeconomicdevelopment(t‐statisticofEco_Dev=13.818)are
positive and significant at the 1 percent level. These results suggest that stronger
accounting environment (via IFRS adoption and greater disclosure index), stronger
investor protection, and higher economic development are all related to lower level of
corruption.
AsTable4reveals,whenPol_InsreplacesAcc_Envinmodel(2),theadjustedR2improves
to87.4%from60.9%inmodel(1). NowPol_Ins(t‐statistic=18.857)ispositivelyrelated
3Allstatisticaltestsinthisstudyarebasedontwo‐tailedtests.
Page | 18
to the corruption perception index and statistically significant at the 1 percent level.
Althoughthesignificanceofinvestorprotection(t‐statisticofInv_Pro=1.759)nowweakens
to 10 percent level,Eco_Dev (t‐statistic =5.480) is still significant at the 1 percent level.
Model(3)incorporatesbothAcc_EnvandPol_InsalongwiththecontrolvariablesInv_Pro
and Eco_Dev. In model (3) results, both Acc_Env and Pol_Ins are positively related to
corruptionperception index, butAcc_Env (t‐statistic =1.872) is only significant at the 10
percent levelwhereasPol_Ins (t‐statistic=18.745) is significant at the1percent level. In
model(3)results,investorprotectionisnotsignificantatconventionallevels,althoughthe
levelofeconomicdevelopmentstill retains itssignificanceat the10percent level. Thus,
Table 4 clearly demonstrates that, although stronger accounting environment has some
negative influence on corruption, the strength of the political institutions is the most
dominantvariableinexplainingcorruptionperception.
(InsertTable4here)
InTable5,wereport theestimatesofmodels(4), (5),and(6) using2SLStechnique.4
Theresultsfrom2SLSarelargelyconsistentwiththoseoftheOLSestimatesreportedin
Table 4. As Table 5 reveals, both Acc_Env (coefficient = 0.173, t‐statistic =3.450) and
Pol_INS (coefficient =0.370, t‐statistic =10.470) have positive coefficients and both are
statistically significant at the 1 percent level. However, larger coefficient and larger t‐
statistic forPol_Ins compared to thoseofAcc_Env suggest that the corruptionperception
indexismuchmoresensitivetothestrengthofpoliticalinstitutionsthanitistoaccounting
environment. Further, positive and statistically significant coefficient on Eco_Dev (t‐
statistic=3.440,p=0.001)areconsistentwithpriorresearchthatdevelopedcountrieshave
4TheHausman(1978)test(χ2=24.33,p=0.0002)confirmsendogeneityinthemodels.
Page | 19
lower levels of corruption (e.g., Kimbro, 2002; Treisman, 2007). In model (5) where
Acc_Env is the dependent variable, the corruption perception index (Low_Corrup) is
positiveandsignificantatthe10percentlevel(t‐statistic=1.790).Thisresultisconsistent
with previous findings (e.g., Kimbro, 2002; Malagueño et al., 2010) that low level of
corruption is associated with improved accounting environment. In model (5), the
strengthofpoliticalinstitutions(Pol_Ins)isnegativebutinsignificant(t‐statistic=‐0.740).
Thus, the strength of political institutions does not directly improve accounting
environment,butpoliticalinstitutionsindirectlyimproveaccountingenvironmentthrough
their effect on corruption. In model (6) where Pol_Ins is the dependent variable, the
corruption perception index (Low_Corrup t‐statistic =17.560) is positive and statistically
significantatthe1percentaaaarriinnnnnnnaaasartgs7xe445ws5confirmingpriorresults
of a strong positive relation between the strength of political institutions and control of
corruption. In sum, results in Table 5 confirm that, after controlling for endogeneity
among control of corruption, political institutions, and accounting environment, the
strengthofpolitical institutionshasthestrongestinfluenceonthecorruptionperception
level.
(InsertTable5here)
6.Robustnesschecks
In this section,we report the resultsof various robustness tests. InTable6, we re‐
estimate the models (1) to (3) on a sample which comprises data from 2002 to 2011.
BecausetheIASBannounceditsfirstprogramoftechnicalprojectstoimprovestandardsin
2001 and a major momentum for global adoption of IFRS standards began with the
European Union’s announcement in 2002 to adopt the IFRS, we modify our sample to
Page | 20
compriseaccountingenvironmentdatasince2002.Resultsbasedonthismodifiedsample
are reported in Table 6. As Table 6 reveals, Acc_Env (t‐statistic =1.309) now loses its
significance in the control of corruption, although Inv_Pro and Eco_Dev are statistically
significantatthe1percentlevel.Resultsofmodels(2)and(3)arequalitativelysimilarto
thosereportedinTable4.InTable6,althoughthesignificancelevelofAcc_Env(t‐statistic
=2.323)intheresultsofmodel(3)improvesfrom10percentlevelinTable4to5percent
levelinTable6,Pol_Ins(t‐statistic=19.139,p<0.001)isstillthemostdominantdriverin
explainingthecorruptionperceptionindex.
(InsertTable6here)
Resultssofarwerebasedonthemeanmeasuresforeachvariable.Nowwere‐specify
models(1)to(3)usingindividualcomponentsineachmeasureofaccountingenvironment
andthestrengthofpoliticalinstitutions.Thesere‐specifiedmodelsareasfollows:
Low_Corrup=α0+α1(Acc_Env)+α2(Rule_Law)+α3(Inv_Pro)+α4(Eco_Dev)+ε(7)
Low_Corrup=α0+α1(Acc_Env)+α2(Press_Freedom)+α3(Inv_Pro)+α4(Eco_Dev)+ε(8)
Low_Corrup=α0+α1(IFRS)+α2(Pol_Ins)+α3(Inv_Pro)+α4(Eco_Dev)+ε(9)
Low_Corrup=α0+α1(Disclosure)+α2(Pol_Ins)+α3(Inv_Pro)+α4(Eco_Dev)+ε(10)
whereallvariablesareasdefinedinTable1.
As Table 7 reveals, inmodel (7)Acc_Env is not significant at conventional levels (t‐
statistic = 0.516) when only rule of law (Rule_Law) is used as the proxy for political
institutions. In model (7), clearly rule of law (Rule_Law) is the dominant variable (t‐
statistic =28.401, p<0.001). In model (8), the variable Acc_Env (t‐statistic =3.020) is
positive and significant at the 1 percent level, but Press_Freedom (t‐statistic =11.625,
p<0.001), the proxy for the strength of political institutions, is the strongest variable in
Page | 21
explaining theperceptionof corruption. Inmodel (9), theadoptionof IFRS (t‐statistic=
0.017, p=0.986) is not at all significant, whereas the strength of political institutions
(Pol_Ins t‐statistic=19.568) is significantat the1percent level. Finally, inmodel (10),
although the disclosure of ownership and financial information (Disclosure: coefficient =
0.030, t‐statistic =2.747), the proxy for accounting environment, is significant at the 1
percentlevel,Pol_Ins(coefficient=0.426,t‐statistic=20.088)issignificantatthe1percent
levelaswell.Overall,resultsinTable7re‐assurethatthestrengthofpoliticalinstitutions
isthemostsignificantvariableinexplainingthecorruptionperceptionindex.
(InsertTable7here)
TobeconsistentwithDiRienzoetal.(2007)andMalagueñoetal.(2010),weincorporate
four new variables related to Hofstede’s (2001) cultural dimensions in alternative
specificationsofthemodels.Theseculturaldimensionsareindividualism,powerdistance,
uncertaintyavoidance,andmasculinity. Asaresultof including thesenewvariables, the
modifiedmodelsareasfollows:
Low_Corrup = β0 + β1(Acc_Env) + β2 (Pol_Ins) + β3 (Inv_Pro) + β4(Eco_Dev) + β5 (Indiv) +ε (11)
Low_Corrup = λ0 + λ1(Acc_Env) + λ2 (Pol_Ins) + λ3 (Inv_Pro) + λ4(Eco_Dev) + λ5 (Pow_Dis) + ε (12)
Low_Corrup = γ0 + γ 1(Acc_Env) + γ 2 (Pol_Ins) + γ 3 (Inv_Pro) + γ 4(Eco_Dev) + γ5 (Un_Avoid) + ε (13)
Low_Corrup = α0 + α 1(Acc_Env) + α 2 (Pol_Ins) + α 3 (Inv_Pro) + α 4(Eco_Dev) + α5 (Mascu) +ε (14)
Low_Corrup = δ0 + δ 1(Acc_Env) + δ 2 (Pol_Ins) + δ 3 (Inv_Pro) + δ 4(Eco_Dev)
+ δ5 (Indiv) + δ6 (Pow_Dis) + δ7 (Un_Avoid) + δ8 (Mascu) + ε (15)
where Indiv is the measure of individualism, Pow_Dis is the measure of large power
distanceinasociety,Un_Avoidisthetoleranceforuncertainty,andMascuisthemeasure
of how masculine a society is as per Hofstede (2001). Variable definitions of Indiv,
Pow_Dis,Un_Avoid, andMascu areprovided inTable1.Allothervariablesareasdefined
Page | 22
earlier.Becauseofmissingdataonculturaldimensions, in thisanalysiswe haveamuch
reduced sample size of 93 countries. As Table 8 reveals, among the newly incorporated
variables, only strong uncertainty avoidance (Un_Avoid) is significant. In model (13),
Un_Avoid(t‐statistic=‐3.031)isnegativeandsignificantatthe1percentlevel.Model(14)
resultssuggestthatmore‐masculinesocietiesaremorecorruptthanless‐masculineones;
thecoefficientofMascu(t‐statistic=‐1.793)isnegativeandsignificant atthe10percent
level. In model (15), where all cultural dimensions are included, only uncertainty
avoidance (t‐statistic = ‐2.880) is significantly negative at the 1 percent level. These
results suggest that countries with strong uncertainty avoidance tend to have higher
corruptionlevelspresumablybecauseinvestorsandindividualsinthesecountriesresolve
uncertainties and delays in administrative procedures by engaging in bribing. Further,
societies that aremoremasculine, and emphasize onmaterial achievement, tend to be
morecorrupt.InTable8,resultsonthekeyvariablesareconsistentwiththosereportedin
previoustables.
(InsertTable8here)
Finally, we incorporate the Big4 market share as a proxy for accounting quality, as
employed by Kimbro (2002) and Malagueño et al. (2010). We find consistent results
(untabulated) for the strength of political institutions and accounting environment. The
strengthofpoliticalinstitutionsisstillthemostdominantvariableinexplainingcorruption
perception.
In sum, all empirical analyses suggest that the strength of political institutions is the
most dominant driver in the control of corruption and the (quality of) accounting
environmentplaysonlysecondaryorsupportiverole. Thus,countrieswishing toreduce
Page | 23
corruption,andpromoteinvestmentsandgrowthintheeconomywillbemuchbetteroff
by strengthening their political institutions. Adoption of higher quality accounting
standards can deliver higher‐quality financial reporting only in the presence of strong
political institutions in the country. Our results indirectly explain themixedevidenceof
financial reporting quality found in cross‐country studies on the adoption of IFRS.
Countries with weak political institutions cannot expect to have improved financial
reportinginarealsensewiththeadoptionofIFRSbecausetheincentivesemanatingfrom
theinstitutionalsettingofacountryaremorefundamentalthanthestandardsthemselves
indeterminingfinancialreportingquality(Ball,Robin&Wu,2003;Ball,2006).AsCampos
etal.(1999)state,“(C)ountriesthatinvestinitiallyininhibitingcorruptionmayultimately
haveastrongerfoundationforsustaininggrowthoverthelongterm”(p.1065).
7.Conclusion
In this study,we investigated the relationbetween the accounting environment in a
countryandperceptionofcorruption,aftercontrollingfortheroleofpoliticalinstitutions
and economic development in the control of corruption. We were motivated by the
paucityofresearchinthisareaandtheprematureconclusionreachedbyMalagueñoetal.
(2010)ontheroleofaccountingincontrollingcorruption.
Wepooled together datarelatedtocorruptionperceptionindex and thestrengthof
political institutions (voice and accountability, and rule of law) from Kaufmann et al.
(2012), and investorprotectionandeconomicdevelopmentdata from theWorldBank
(2010).Wemeasuredaccountingenvironmentalongtwodimensions:(1)whetherornot
a country has adopted the IFRS; and (2) the extent to which investors are protected
Page | 24
through disclosure of ownership and financial information. We found that, although
accounting environment has some positive effect in the control of corruption, its role is
relatively minor and secondary to the effect of political institutions. Our results are
robust to alternative specifications of our models. More importantly, our results are
consistent whether we estimated our models using OLS or 2SLS to account for the
endogenousrelationamongpoliticalinstitutions,corruption,andaccountingenvironment.
OurresultsalsoholdwhenweincorporateHofstede’s(2001)culturaldimensionsinour
models.
Our results suggest that the proxies used inMalagueño et al. (2010)may have been
inadequate in capturing the strength of political institutions in a country, and thus, the
optimism they placed on accounting environment in the control of corruption is
unwarranted. In all our estimations, the strength of political institutions, as proxied by
voiceandaccountabilityandruleoflaw,appearstohavethemostsignificantpositiveeffect
onthecorruptionperceptionindex.Thus,strengtheningthepoliticalinstitutionswillhave
thelargestimpactincontrollingcorruptioninacountry.
Ourresultshave implications for thecountries thathaveadopted IFRS inone formor
another.Becauseinterpretationofaccountingstandardsultimatelyrestswiththeauditors,
courts, and judges of a country, those countries that have the strongest political
institutionsstandtobenefitmostfromtheadoptionofIFRS.
Althoughourresultsarestrongandconsistent, theyneedtobe interpretedwithsome
caution. First, like most cross‐country studies, we pooled data from different sources
whichcollectanddisseminatedatafordifferentpurposes.Thus,constructvalidityofsome
Page | 25
of our proxiesmight be an issue. Second, our results could be sensitive to alternative
proxies for the variables used in the study. Nevertheless, we provide some strong
evidenceoftheroleofaccountingenvironmentincontrollingcorruption.
Page | 26
References
Ades, A., & Di Tella, R. (1999). Rents, Competition and Corruption, American Economic
Review,89(4),982‐93.
Ahmed, S, A., Neel, M., & Wang, D. (2013). Does Mandatory Adoption of IFRS Improve
Accounting Quality? Preliminary Evidence, Contemporary Accounting Research,
availableatdoi:10.1111/j.1911-3846.2012.01193.x
Aidt, T.K. (2003). Economic analysis of corruption:A survey,TheEconomic Journal, 113
(491),F632‐F652.
Alesina,A.,&Rodrik,D.,(1994).Distributivepoliticsandeconomicgrowth,TheQuarterly
JournalofEconomics109(2),465‐490.
Ali, A. M., & Isee, H. S. (2003). Determinants of economic corruption: A cross‐country
comparison,CatoJournal22(3),449‐466.
Ashbaugh, H., & Pincus, M. (2001). Domestic Accounting Standards, International
Accounting Standards and the Predictability of Earnings, Journal of Accounting
Research,39(3),417‐434.
Bahmani‐Oskooee,M.,&Nasir,A.(2002).Corruption,lawandorder,bureaucracy,andreal
exchangerate,EconomicDevelopmentandCulturalChange50(4),1021‐1028.
Ball,R.,Robin,A.,&Wu,J.S.(2003).IncentivesversusStandards:PropertiesofAccounting
Income in Four East Asian Countries, Journal ofAccounting and Economics, 36(1‐3),
235‐270.
Barth, M., Landsman, W., & Lang, M. (2008). International Accounting Standards and
AccountingQuality,JournalofAccountingResearch,46(3),467‐498.
Blackburn,K.,Bose,N.,&Haque,M.E.(2006).Theincidenceandpersistenceofcorruption
in economic development, Journal ofEconomicDynamics& Control, 30 (12), 2447‐
2467.
Blackburn, K., N. Bose, & Haque, M. E. (2010). Endogenous corruption in economic
development,JournalofEconomicStudies37(1),4‐25.
Page | 27
Bushman, R. M., Piotroski, J. D., & Smith, A. J. (2004). What Determines Corporate
Transparency?JournalofAccountingResearch,42(2),207‐252.
Campos, J. E., Lien, D. & Pradhan, S. (1999). The impact of corruption on investment:
Predictabilitymatters,WorldDevelopment27(6),1059‐1067.
Deloitte’s IAS Plus Website. (2012). Use of IFRS by Jurisdiction. Retrieved from
http://www.iasplus.com/Plone/en/resources/use‐of‐ifrs
Ding, Y., Hope, O.K., Jeajean, T., & Stolowy, H. (2007). Difference between Domestic
AccountingStandardsandIAS:Measurement,DeterminantandImplications,Journalof
AccountingandPublicPolicy,26(1),1‐38.
DiRienzo, E., Das, J., Cort, K.T., & Burbridge, J. (2007). Corruption and the Role of
Information,JournalofInternationalBusinessStudies,38(2),320‐332.
Ehrlich, I.,&Lui,F.T. (1999)Bureaucraticcorruptionandendogenouseconomicgrowth,
JournalofPoliticalEconomy107(S6),S270‐S293.
Everett, J., Neu, D., & Rahaman A. S. (2007). Accounting and the Global Fight against
Corruption,Accounting,OrganizationsandSociety,32(6),513–542.
Fisman,R.,& Svensson, J. (2007).Are corruption and taxation reallyharmful to growth?
Firmlevelevidence,JournalofDevelopmentEconomics,83(1),63‐75.
Ghoshal,S.,&Moran,P.(2005).TowardsaGoodTheoryofManagement,inJ.Birkinshawan
&G.Piramal(eds.)SumantraGhoshalonManagement:AForceforGood,UpperSaddle
River,NJ:FinancialTimes/PrenticeHall,1‐27.
Habib, M., & Zurawicki, L. (2002). Corruption and Foreign Direct Investment, Journal of
InternationalBusinessStudies,33(2),291‐307.
Hausman, J. A. (1978). Specification tests in econometrics, Econometrica, 46(6), 1251–
1271.
Hofstede, G. H. (2001). Culture's Consequences: ComparingValues,Behaviors, Institutions,
andOrganizationsacrossNations.,ThousandOaks,CA:SagePublications.
Hope,O.K.,Jin,J.,&Kang,T.(2006).EmpiricalEvidencefromJurisdictionsthatAdoptIFRS.
JournalofInternationalAccountingResearch,5(2),1‐20.
Page | 28
Houqe, M.N., Monem, R., & van Zijl, T. (2012a).Government Quality, Auditor Choice and
AdoptionofIFRS:aCrossCountryAnalysis,AdvancesinAccounting,28(2),307‐316.
Houqe,M.N.,vanZijl,T.,Dunstan,K.,&Karim,W.(2012b).TheEffectofIFRSAdoptionand
InvestorProtectiononEarningsQualityaroundtheWorld,TheInternationalJournalof
Accounting,47(3),333‐355.
Jain,A.(2001).Corruption:areview,JournalofEconomicSurveys,15(1),71‐121.
Jeanjean,T.,&Stolowy,H.(2008).Doaccountingstandardsmatter?Anexploratoryanalysis
ofearningsmanagementbeforeandafterIFRSadoption,JournalofAccounting&Public
Policy,27(6),480‐494.
Kaufmann,D.,&Wei,S.J.(1999).Does‘greasemoney’speedupthewheelsofcommerce?
NBERWorkingPaperSeries.
Kaufmann, D., Kraay, A., &Mastruzzi,M. (2012). TheWorldwide Governance Indicators,
Availableathttp://info.worldbank.org/governance/wgi/index.asp
Kehoe,W.J.(1998).TheEnvironmentofEthicsinGlobalBusiness,JournalofBusinessandBehaviouralBehavioralScience,2(1),47‐56.
Kimbro, M.B. (2002). A Cross‐Country Empirical Investigation of Corruption and its
RelationtoEconomic,Cultural,andMonitoringInstitutions:anexaminationoftherole
of Accounting and Financial Statements Quality, Journal of Accounting Auditing and
Finance,17(4),325‐75.
Knack, S. (2007). Measuring Corruption: A Critique of Indicators in Eastern Europe and
CentralAsia,JournalofPublicPolicy,27(3),255–291.
La Porta, R., Lopez‐de‐Silanes, F., & Shleifer, A.S, & Vishny, R. (1998). Law and Finance,
JournalofPoliticalEconomy,106(6),1113‐1155.
Lambert‐Mogiliansky, A. (2002). Why firms pay occasional bribes: The connection
economy,EuropeanJournalofPoliticalEconomy18(1),47‐60.
Lambsdorff,J.G.(2006).ConsequencesandCausesofCorruption:WhatDoWeKnowFrom
aCross‐SectionofCountries?InInternationalHandbookontheEconomicsofCorruption,
Rose‐Ackerman(ed.),EdwardElgar,Cheltenham,UK.
Page | 29
Lederman, D., Loayza, N., & Soares, R. S. (2005). Accountability and Corruption Political
InstitutionsMatter,Economics&Politics,17(1),1–35.
Malagueno,R.,Albrecht,C.,Ainge,C.,&Stephens,N.(2010).AccountingandCorruption:A
Cross‐CountryAnalysis.JournalofMoneyLaunderingControl,13(4),372‐393.
Mauro,P.(1995).CorruptionandGrowth.QuarterlyJournalofEconomics,110(3),681‐712.
Mauro,P.(1997).TheEffectofCorruptiononGrowth,InvestmentandGrowthExpenditure:
a Cross Country Analysis. In K.A Elliott (Ed.),Corruption inGlobalEconomy, 83‐107,
Washington,DC:InstituteforInternationalEconomics.
Mo, P.H. (2001). Corruption and Economic Growth, Journal of Comparative Economics,
29(1),66‐79.
Paldam,M. (2002). The cross‐countrypattern of corruption: Economics, culture, and the
seesawdynamics,EuropeanJournalofPoliticalEconomy18(2),215‐240.
Rock,M.T.,&Bonnett,H.(2004). Thecomparativepoliticsofcorruption:Accountingfor
theEastAsianparadox in empirical studies of corruption, growth and investment,
WorldDevelopment32(6),999‐1017.
Rose‐Ackerman,S. (2003). Corruption in editedbyC.K.Rowley andF. Schneider (Ed(s),
TheEncyclopaediaofPublicChoiceCorruption,67‐76,KluwerAcademicPublishers
Sandholtz, W., & Koetzle, W. (2000). Accounting for Corruption: Economic Structure,
Democracy,andTrade.InternationalStudiesQuarterly,44(1),31‐50.
Shleifer,A.,&Vishny,R. (1993).Corruption.TheQuarterly JournalofEconomics, 108 (3),
599‐617.
Smarzynska, B. K., & Wei, S. J. (2000). Corruption and composition of foreign direct
investment:Firm‐levelevidence,NBERWorkingPaperSeries.
Soderstrom, N. S., & Sun, K. J. (2007). IFRS adoption and accounting quality: A review,
EuropeanAccountingReview,16(4),605‐702.
Tanzi, V., (1998). Corruption around the world: Consequences, scope, and cures, Staff
Papers–InternationalMonetaryFund45(4),559‐594.
Page | 30
The World Bank. (2012). Data and Indicators 2010, Retrieved from
http://data.worldbank.org/indicator
The World Bank. (2010). Data and Indicators 2010, Retrieved from
http://data.worldbank.org/indicator
Treisman,D.(2007).Whathavewelearnedaboutthecausesofcorruptionfromtenyears
ofcross‐nationalempiricalresearch?AnnualReviewofPoliticalScience,10,211‐244.
Wei, S. J. (2000), How taxing is corruption on international investors? The Review of
EconomicsandStatistics,82(1),1‐11.
World Economic Forum. (2010). The Global Competitiveness Report 2010–2011,
Switzerland.
Wu, X. (2005a). Corporate Governance and Corruption: A Cross‐Country Analysis,
Governance:An International JournalofPolicy,Administrationand Institutions,18 (2),
151‐170.
Wu, X. (2005b). Firm Accounting Practices, Accounting Reforms and Corruption in Asia,
PolicyandSociety,24(3),53‐78.
Zarb, B.J. (2008). Do Accounting Regulation, Transparency and the Propensity to Bribe
affect the Perception of Corruption in Developed Countries? Journal of International
FinanceandEconomics,8(4),164.
Zhao,J.H.,Kim,S.H.,&Du,J.(2003).TheImpactofCorruptionandTransparencyonForeign
DirectInvestment:anEmpiricalAnalysis,ManagementInternationalReview,43(1),41‐
62.
Page | 31
Table1:Descriptionofvariablesandsources
Variable Measure Descriptionofvariable DatasourceDependentvariableCorruption Controlofcorruption
(Low_Corrup)Averageofthecontrolofcorruptionindex for the years 1996, 1998,2000,2002,2003,2004,2005,2006,2007, 2008, 2009, 2010, and 2011.This measure reflects a perceptionof theextent towhichpublicpoweris exercised for private gain,includingbothpettyandgrandformofcorruption,aswell“captureofthestatebyelitesandprivate interest”.It ranges from approximately ‐1.5981 to 2.4411, with a higherscoreindicatingleastcorruptregimeandvice–versa.
Kaufmann etal.,(2012)
IndependentvariablesAccountingEnvironment
Acc_Env1. IFRS2.ExtentofDisclosureIndex
Aggregatescoreoftwomeasures:Adummyvariabletakesthevalueofone (1) if a country has adoptedIFRSandzero(0)otherwise.Disclosure index measures theextent to which investors areprotected through disclosure ofownership and financialinformation. The index ranges from0 to 10, with higher valuesindicatinggreaterdisclosure.
DeloitteIASPluswebsite(2012)World Bank(2012)
Politicalinstitutions
Pol_Ins1.VoiceandAccountability
Aggregatescoreoftwomeasures:Average of the voice andaccountability index for the years1996,1998,2000,2002,2003,2004,2005,2006,2007,2008,2009,2010,and2011.Itmeasures“theextenttowhich a country’s citizens are ableto participate in selecting theirgovernment, as well as freedom ofexpression, freedom of associationand a free media”. It ranges from ‐1.88 to 1.60, with higher scoresindicating greater voice and
Kaufmann etal.,(2012)
Page | 32
accountabilityaccessandvice‐versa. 2.RuleofLaw Averageof theruleof law index for
the years 1996, 1998, 2000, 2002,2003,2004,2005,2006,2007,2008,2009, 2010, and 2011.. Itmeasuresthe extent to which agents haveconfidenceinandabidebytherulesof society, and in particular thequalityof contractenforcement, thepolice,andthecourts,aswellasthelikelihood of crime and violence. Itranges from ‐1.82 to 1.94, withhigher scores indicating strong ruleoflawandvice‐versa.
Kaufmann etal.,(2012)
ControlvariablesInvestorProtection
Strength of InvestorProtection(Inv_Pro)
Strengthof investorprotection indexmeasures the degree to whichcorporate laws protect the rights ofinvestors,borrowersandlendersandthus facilitate lending. The indexranges from 0 to 10, with higherscores indicating that these laws arebetter designed to protect investorinterest.
World Bank(2012)
EconomicDevelopment
Natural logarithm ofGDPpercapita(US$)(Eco_Dev)
GDP per capita is gross domesticproduct divided by midyearpopulation. GDP is the sum of grossvalueaddedbyallresidentproducersin the economy plus any producttaxes and minus any subsidies notincludedinthevalueoftheproducts.It is calculated without makingdeductions for depreciation offabricated assets or depletion anddegradation of natural resources.DataareincurrentU.S.dollars.
World Bank(2010)
Culture POW_Dis
Power distance: It measures theresponse of people to inequality andtheextenttowhichtheless‐powerfulmembers expect, accept, or evenprefer the fact that power isdistributed unequally. Cultures withan unequal distribution of power
Hofstede(2001)
Page | 33
IndivUn_Avoid
Mascu
tend to discourage questioningauthority.
Individualism: It refers to the extentthat individuals are integrated intogroups.Itreflectsthatcountrieswithhigh level of individualism places ahigher value of individualachievement and responsibility.Individualistic societies have agreater tolerance of diversity anddifferences of opinion. Theoppositeofindividualismiscollectivismwheregroup or societal norms takeprecedenceoverindividualviews.
Uncertainty avoidance: It measuresthesociety’s toleranceofuncertaintyor unknown situation. Societies thathave high uncertainty avoidance arethose in which people feeluncomfortable in unpredictedsituations which result tounwillingnesstochallengeauthority.
Masculinity index: It measures theextenttowhichasocietyemphasizescompetition and wealth acquisitionover relationship with others andquality of life. Japan is the mostmasculine societywith a score of 95and Sweden is the most femininesocietywithascoreof5.
Hofstede(2001)Hofstede(2001)Hofstede(2001)
Page | 34
Table2:Descriptivestatistics
Variables Low_Corrup Acc_Env Pol_Ins Inv_Pro Eco_DevMean ‐0.046 6.301 ‐0.196 4.536 8.527Median ‐0.304 6.000 ‐0.518 53.000 8.512SD 0.996 2.499 1.836 2.485 1.513Minimum ‐1.599 0.000 ‐3.299 1.000 5.440Maximum 2.441 11.000 3.516 9.000 11.650N 166 166 166 166 166
AllvariabledefinitionsappearinTable1.
Table3:Correlationmatrix
Variables Low_Corrup Acc_Env Pol_Ins Inv_Pro Eco_DevLow_Corrup 1 0.223***
(0.004)0.911***(<0.001)
0.351***(<0.001)
0.686***(<0.001)
Acc_Env 0.260***(0.001)
1 0.201***(0.009)
0.160**(0.039)
0.191**(0.014)
Pol_Ins 0.923***(<0.001)
0.215***(0.005)
1 0.332***(<0.001)
0.652***(<0.001)
Inv_Pro 0.345***(<0.001)
0.154**(0.047)
0.332***(<0.001)
1 0.186**(0.017)
Eco_Dev 0.752***(<0.001)
0.184**(0.017)
0.699***(<0.001)
0.197**(0.011)
1
AllvariabledefinitionsappearinTable1.
***,**,*Correlationssignificantat0.01,0.05,and0.10levels,(two‐tailedtests),respectively.
Page | 35
Table4
OLSregressionanalysisofcorruption(dependentvariableislowcorruption)
Low_Corrup=α0+α1(Acc_Env)+α2(Inv_Pro)+α3(Eco_Dev)+ε(1)
Low_Corrup=ά0+ά1(Pol_Ins)+ά2(Inv_Pro)+ά3(Eco_Dev)+ε(2)
Low_Corrup=ψ0+ψ1(Acc_Env+ψ2(Pol_Ins)+ψ3(Inv_Pro)+ψ4(Eco_Dev)+ε(3)
Independentvariables Model1Estimate(p‐value)t‐statistic
Model2Estimate(p‐value)t‐statistic
Model3Estimate(p‐value)t‐statistic
Intercept ‐1.078***(<0.001)‐7.223
‐0.189***(0.006)‐2.798
‐0.311***(0.001)‐3.327
Acc_Env 0.041**(0.043)2.044
0.021*(0.063)1.872
Pol_Ins 0.411***(<0.001)18.857
0.407***(<0.001)18.745
Inv_Pro 0.077***(<0.001)3.834
0.021*(0.080)1.759
0.019(0.113)1.592
Eco_Dev 1.543***(<0.001)13.818
0.470***(<0.001)5.480
0.462***(<0.001)5.415
Adj.R2 0.609 0.874 0.876N 166 166 166
AllvariabledefinitionsappearinTable1.a. ***, **, * indicate statistical significance at the 0.01, 0.05, and 0.10 levels (two‐tailed tests),respectively.b.Dependentvariableiscontrolofcorruption.
Page | 36
Table5
Simultaneousequationanalysis(2SLS)forthecorruption,politicalinstitutionandaccountingquality
Low_Corrup=α0+α1(Acc_Env)+α2(Pol_Ins)+α3(Inv_Pro)+α4(Eco_Dev)+ε(4)
Acc_Env=ψ0+ψ1(Low_Corrup)+ψ2(Pol_Ins)+ψ3(Inv_Pro)+ψ4(Eco_Dev)+ε(5)
Pol_Ins=ά0+ά1(Acc_Env)+ά2(Inv_Pro)+ά3(Eco_Dev)+ε(6)
Independentvariables
Model(4)Dep.Var.=Low_Corrup
Estimate(p‐value)t‐statistic
Model (5)Dep.Var.=Acc_Env
Estimate(p‐value)t‐statistic
Model(6)Dep.Var.=Pol_Ins
Estimate(p‐value)t‐statistic
Intercept ‐0.1.248***(<0.001)‐3.890
‐6.295***(<0.001)1.790
‐0.129*(0.380)‐0.880
Low_Corrup 1.060*(0.074)1.790
1.699***(<0.001)17.560
Acc_Env 0.1727***(0.001)3.450
Pol_Ins 0.370***(<0.001)10.470
‐0.2150(0.458)‐0.740
Inv_Pro 0.0062***(0.736)0.340
0.046**(0.594)0.530
0.007(0.776)
0.280Eco_Dev 0.472***
(0.001)3.440
‐0.279***(0.699)‐0.390
‐0.016(0.940)‐0.080
Durbinχ2
160.37***
Wu‐HausmanF‐statistic166.740***
R2 0.756 0.074 0.856N 166 166 166
AllvariabledefinitionsappearinTable1.a. ***, **, * indicate statistical significance at the 0.01, 0.05, and 0.10 levels (two‐tailed tests),respectively.
Page | 37
Table6
OLSregressionanalysisofcorruption:Re‐estimationofmodels(1)to(3)usingdatafrom2002to2011
Low_Corrup=α0+α1(Acc_Env)+α2(Inv_Pro)+α3(Eco_Dev)+ε(1)
Low_Corrup=ά0+ά1(Pol_Ins)+ά2(Inv_Pro)+ά3(Eco_Dev)+ε(2)
Low_Corrup=ψ0+ψ1(Acc_Env+ψ2(Pol_Ins)+ψ3(Inv_Pro)+ψ4(Eco_Dev)+ε(3)
Independentvariables Model (1)Estimate(p‐value)t‐statistic
Model(2)Estimate(p‐value)t‐statistic
Model(3)Estimate(p‐value)t‐statistic
Intercept ‐0.946***(<0.001)‐6.852
‐0.258***(<0.001)‐4.069
‐0.387***(<0.001)‐4.621
Acc_Env 0.025(0.192)1.309
0.025**(0.021)2.323
Pol_Ins 0.394***(<0.001)18.900
0.394***(<0.001)19.139
Inv_Pro 0.078***(<0.001)4.055
0.027**(0.018)2.386
0.023**(0.040)2.069
Eco_Dev 1.501***(<0.001)13.443
0.525***(<0.001)6.353
0.498***(<0.001)6.038
Adj.R2 0.580 0.861 0.864N 166 166 166
AllvariabledefinitionsappearinTable1.a. ***, **, * indicate statistical significance at the 0.01, 0.05, and 0.10 levels (two‐tailed tests),respectively.b.Dependentvariableiscontrolofcorruption.
Page | 38
Table7
OLSregressionanalysisofcorruption
Low_Corrup=α0+α1(Rule_Law)+α2(Acc_Env)+α3(Inv_Pro)+α4(Eco_Dev)+ε(7)
Low_Corrup=α0+α1(Press_Freedom)+α2(Acc_Env)+α3(Inv_Pro)+α4(Eco_Dev)+ε(8)
Low_Corrup=α0+α1(Pol_Ins)+α2(IFRS)+α3(Inv_Pro)+α4(Eco_Dev)+ε(9)
Low_Corrup=α0+α1(Pol_Ins)+α2(Disclosure)+α3(Inv_Pro)+α4(Eco_Dev)+ε(10)
Variables Model(7)Estimate(p‐value)t‐statistic
Model (8)Estimate(p‐value)t‐statistic
Model ( 9)Estimate(p‐value)t‐statistic
Model(10)Estimate(p‐value)t‐statistic
Intercept 0.073(0.387)(0.867)
‐0.413***(0.004)‐2.885
0.084***(0.386)0.869
‐0.067(0.528)‐0.632
Acc_Env 0.005(0.516)0.651
0.044***(0.003)3.020
IFRS 0.001(0.986)0.017
Disclosure 0.030***(0.007)2.747
Pol_Ins 0.426***(<0.001)19.568
0.426***(<0.001)20.088
Rule_Law 0.941***(<0.001)28.401
Press_Freedom 0.571***(<0.001)11.625
Inv_Pro 0.020*(0.054)1.936
0.021(0.269)1.108
0.036**(0.014)2.492
0.037**(0.010)2.611
Eco_Dev 0.134*(0.052)1.954
0.950***(<0.001)9.554
0.500***(<0.001)5.947
0.473***(0.007)2.747
Adj.R2 0.921 0.860 0.860 0.866N 166 166 166 166
AllvariabledefinitionsappearinTable1.a. ***, **, * indicate statistical significance at the 0.01, 0.05, and 0.10 levels (two‐tailed tests),respectively.b.Dependentvariableiscontrolofcorruption.
Page | 39
Table8
OLSregressionanalysisofcorruptionwithHofstede’s(2001)measuresofcultureLow_Corrup = β0 + β1(Acc_Env) + β2 (Pol_Ins) + β3 (Inv_Pro) + β4(Eco_Dev) + β5 (Indiv) + ε (11)
Low_Corrup = λ0 + λ1(Acc_Env) + λ2 (Pol_Ins) + λ3 (Inv_Pro) + λ4(Eco_Dev) + λ5 (Pow_Dis) + ε (12)
Low_Corrup = γ0 + γ 1(Acc_Env) + γ 2 (Pol_Ins) + γ 3 (Inv_Pro) + γ 4(Eco_Dev) + γ5 (Un_Avoid) + ε (13)
Low_Corrup = α0 + α 1(Acc_Env) + α 2 (Pol_Ins) + α 3 (Inv_Pro) + α 4(Eco_Dev) + α5 (Mascu) + ε (14)
Low_Corrup = δ0 + δ 1(Acc_Env) + δ 2 (Pol_Ins) + δ 3 (Inv_Pro) + δ 4(Eco_Dev)
+ δ5 (Indiv) + δ6 (Pow_Dis) + δ7 (Un_Avoid) + δ8 (Mascu) + ε (15)
Variables Model(11)Estimate(p‐value)t‐statistic
Model (12)Estimate(p‐value)t‐statistic
Model (13)Estimate(p‐value)t‐statistic
Model (14)Estimate(p‐value)t‐statistic
Model(15)Estimate(p‐value)t‐statistic
Intercept ‐0.178(0.264)‐1.124
0.059(0.784)0.275
0.278(0.159)1.422
0.008(0.964)0.045
0.577**(0.046)2.026
Pol_Ins 0.409***(<0.001)11.731
0.391***(<0.001)10.658
0.416***(<0.001)13.518
0.412***(<0.001)12.881
0.400***(<0.001)10.820
Acc_Env 0.023(0.129)1.532
0.026*(0.079)1.815
0.016(0.263)1.126
0.032**(0.033)2.164
0.024(0.103)1.650
Inv_Pro 0.017(0.378)0.886
0.017(0.370)0.901
0.021(0.233)1.202
0.013(0.481)0.707
0.017(0.322)0.996
Eco_Dev 0.565***(<0.001)4.579
0.571***(<0.001)4.819
0.595***(<0.001)5.222
0.565***(<0.001)4.793
0.576***(<0.001)4.990
Indiv 0.001(0.692)0.398
‐0.000(0.903)‐0.122
Pow_Dis ‐0.003(0.204)‐1.281
‐0.003(0.369)‐0.904
Un_Avoid ‐0.006***(0.003)‐3.031
‐0.005***(0.005)‐2.880
Mascu ‐0.005*(0.077)‐1.793
‐0.004(0.101)‐1.659
AdjR2 0.891 0.899 0.901 0.901 0.910N 93 93 93 93 93
AllvariabledefinitionsappearinTable1.
Page | 40
a. ***, **, * indicate statistical significance at the 0.01, 0.05, and 0.10 levels (two‐tailed tests),respectively.b.Dependentvariableiscontrolofcorruption.