ContentslistsavailableatScienceDirect
International Journal of Hospitality Management
j o ur na l h o me pa g e :w w w . e l s e v i e r . c o m / l o c a t e / i j h o s m a n
The influence of firm characteristics on earnings quality
Laura Parte-Esteban
a,∗, Cristina Ferrer García
baDepartmentofFinanceandAccounting,FacultyofEconomics,UniversidadNacionaldeEducaciónaDistancia(UNED),Madrid,Spain
bDepartmentofAccountingandFinance,FacultyofSocialSciencesandHumanities,UniversityofZaragoza,Teruel,Spain
a r t i c l e i n f o
Keywords:
Earningsquality Spanishhotelindustry Businessstrategy Ownershipstructure Auditfunction
a b s t r a c t
ThisstudyexaminestheeffectoffirmcharacteristicsonearningsqualityusingasampleofSpanish hotelfirmsduringtheperiod2000–2011.First,webuildamultidimensionalmeasureofearningsquality includingattributesaspersistence,predictability,variabilityandearningssmoothing.Second,weexam- inevariablesthatarepotentiallyassociatedwithearningsqualityandsorttheseintofivecategories:
marketvariables,businessstrategy,ownershipstructure,auditfunctionandcontrolvariables.Thefind- ingssuggestthattheinternationalization,thelocation,theownershipstructureandtheauditfunction influenceearningsqualityinhotelfirms.Thispaperpresentsmanagerialimplicationsforprofessionals, usersoffinancialinformation,andacademics.
©2014PublishedbyElsevierLtd.
1. Introduction
Financialreportsaretheprimarysourceofpubliclyavailable informationaboutacompany’sfinancialpositionandperformance.
Theearningsfigureisthemostimportantfirmindicatoraspro- videdusefulinformationaboutthefirm’svaluationandarehighly influentialindecision-making(SchipperandVincent,2003;Francis etal.,2004).
Earningsmeasuredasoccupancyrate,returnonassets,return onequity,stockreturn,hasbeenwidelyusedasanindicatorfor businessperformanceinhospitalityliterature(see,e.g.Kim and Gu,2005;Chen,2010;TurnerandGuilding,2011;Guilletetal., 2012;Chen et al.,2012; Xiaoet al.,2012).Macroenvironment factorsand company-specificfactorshasbeenextensivelystud- iedaskeyvariablesthatcanhelptoenhancetheearnings(see,e.g.
ÁlvarezGiletal.,2001;Chen,2010).Empiricalstudieshavedemon- stratedthatvariablessuchasthegrowthrateoftotalforeigntourist arrivals,financialcrises, naturaldisasters, political events, loca- tion,internationalization,brandimage,etc.,seemtoaffecttourism firms’strategicorientationaswellastheirperformance.Itisalso notedthatthecorporategovernanceandtheownershipstructure influenceonhotelperformance(BradyandConlin,2004;Perrigot etal.,2009;Turnerand Guilding,2011;Chen etal.,2012;Xiao etal.,2012).Nonetheless,althoughseveralauthorshavestudied accountingquality in thehospitality and tourism literature,no worktodatehasfocusedonearningsqualitymeasuredconsidering
∗Correspondingauthor.Tel.:+34669083415.
E-mailaddresses:[email protected],[email protected] (L.Parte-Esteban),[email protected](C.FerrerGarcía).
asetofattributesandconsequentlythisstudyrepresentsanimpor- tantstepinthehospitalityfield.
Becauseearningsarewidelyusedinvariousscenarios,thequal- ity of reported earnings and theirusefulness has attractedthe attentionofacademics,professionalsandstandardsetters.How- ever, much of the empirical literature in this area focuses on publiclyheldcompaniescomparedtotheirprivatelyheldcounter- parts.Thegreaterdataavailabilityforpublicfirmslikelyexplains thisdiscrepancy.Morrisonetal.(2010)analyzethekeycontrib- utionstosmalltourismbusinessresearchconcludingthatthisfield isarichandfertilesiteforresearchactivity.
Combiningtwo literatures—earningsquality in privatefirms and the hospitality field—we develop our hypothesis. Earnings qualityinprivatefirmshasgarneredconsiderableacademicand publicinterest(Alietal.,2007;Yang,2010;Hopeetal.,2013).The determinantsthatengageinmoreearningsqualityhaveattracted theattentionofacademics,professionalsandstandardsetters.Most previousresearchhasfocusedonmonitoringmechanismssuchas theownershipstructure,theconcentrationofownershipasdrivers ofearningsquality(see,e.g.,Wang,2006;Alietal.,2007;Hopeetal., 2013).Additionally,theeffectsoftaxcompliance(Burgstahleretal., 2006;VanTendelooandVanstralaen,2008;KosiandValentincic, 2013)and debt covenants (Watts and Zimmerman, 1986; Kosi andValentincic,2013)arealsofactorsdirectlylinkedtoearnings quality.
Inhospitalityfield,severalauthorshavefocusedonthefinan- cialreportingand theirquality. Forexample, Jeonet al.(2004) focusonpersistenceofabnormalearningstoanalyzethequality ofearningsinhotelcompanies.TurnerandGuilding(2011)explain themanager’sincentivestouseaccountingchoices(capitalization orexpensingofassetrelatedexpenditures)toaltertheearnings http://dx.doi.org/10.1016/j.ijhm.2014.06.008
0278-4319/©2014PublishedbyElsevierLtd.
figure.ParteEstebanandSuchDevesa(2011a,b)focusonreported earningsanddemonstratetheimportancetoreachcertainearnings benchmarks(smallprofitsagainstsmalllosses).Allofthesestud- ieshaveusedindividualearningsqualitymeasurestodemonstrate theirhypothesisandourstudyincludesearningsqualitymeasure- mentbasedonasetofattributes.
Theobjectiveofthispaperistoanalyzetheearningsquality anditsdeterminants.First,wedevelopamultidimensionalconcept ofearningsqualitythroughfourearningsattributes:persistence, predictability,smoothingandvariability.Second,weexplorethe determinants of earnings quality. We explicitly consider a set ofvariables demonstratedtoaffectearningsqualityin previous researchsuchasleverage,sizeandownershipstructure(Francis et al., 2004; Gaio, 2010) and re-examine the evidence using a sampleofhotelfirmsandamultidimensionalconceptofearnings quality.Moreover,weexaminetheinfluenceofvariablessuchas thefirm’sbusinessmodel(location,internationalizationorprimary activityofthecompany)andmacroeconomicsconditions(financial crisis)thatremainopenquestionsinthefieldofearningsquality.
Ourresearchattemptstofillthisgapintheliterature.Thatis,we exploretheearningsqualitymeasuredasamultidimensionalcon- ceptinasampleofhotelfirmsandproposeasetofdeterminants thatpotentiallyinfluenceonearningsquality.Ourempiricaldesign providesevidencefortherelativeimportanceofindividualfactors andtheirinteractioninshapingreportedearnings.
Wefocusonthetourismindustrybecauseitisafundamental sectoroftheSpanisheconomy.Thecontributionofthesectorto GrossDomesticProduct(GDP)wasapproximately15.2%in2012.
Spainreceivesthesecondmosttourismrevenueintheworldand themostrevenue inEurope(OMT,2013).Specifically,thehotel sub-sectorisconsideredoneofthemostimportantindustriesin theSpanisheconomy.
Oursampleincludesallhotelsfirmsreportingfinancialstate- mentsovertheperiod2000–2011.Inthefirststage,wecompute earningsqualityforeachfirmthroughfourindividualmeasures:
persistence,predictability,smoothnessandvariability. Then,we createanaggregateearningsqualitymeasureforeachfirmbyaver- agingthefourindividualmeasures.Higherrankingsindicatehigher levelsofearningsquality.Inthesecondstage,wedefineasetofvari- ablesthatpotentiallyinfluenceearningsquality.Toincreasethe robustnessoftheanalysis,wedividethefactorsintothefollowing groups:marketvariables,businessstrategy,ownershipstructure, auditfunctionandcontrolvariables.Weanalyzethesefactorsindi- viduallyandthenexaminethemtogether.
Theresultsindicatethatauditfunctionandsomefactorsrelated tobusinessstrategyarestrongdriversofearningsqualityinhotels firms.Wealsofindthatlargerfirmshavehigherearningsquality rankings,whilefirmswithhighersalesvolatility,greateroperat- ingcashflowvolatility,andahigherincidenceoflosseshavelower earningsqualityrankings.Theevidencepresentsstrategicimplica- tionsforhoteliers,accountingstandardsettersandacademics.
Ourcontributiontotheliteratureistoanalyzeadditionaldeter- minantsoftheearningsqualitytendenciesofprivatefirms.Because themajorityofcompaniesintheglobaleconomyareprivateand family-controlledfirms(LaPortaetal.,1999),ourstudyextends previousresearchontheearningsqualityofprivatefirmsandtheir determinantsinseveralways.Inparticular,thisstudyshedslight onaseminalpaperinthehospitalityindustrybyJeonetal.(2004) thatexaminestheearningsqualityofKoreanhotelfirmsusingthe persistenceofearnings.Thisstudysupportstheconceptualization ofearningsqualityasamultidimensionalmeasure.
Furthermore,thisstudyextendstheliteratureonearningsqual- ityin privatefirmsand thehotel industry byincluding several determinantsofearningsquality.Previousstudiestypicallyfocus ononevariablesuchasmanagerialstyle(Chenetal.,2012;Xiao etal.,2012),thelinkbetweenperformanceandCEOcompensation
(KimandGu,2005;Barberetal.,2006;Guilletetal.,2012),fac- tors affectingsystematic risk for thehotel industry(Kim etal., 2012),theeffectofthefinancialcrisisonthehotelperformance (Chaston,2012;Alonso-AlmeidaandBremser,2013),amongoth- ers.Inthisstudy,however,weincludegeneralvariablespreviously documentedtoinfluenceearningsquality(Francisetal.,2004;Gaio, 2010)aswellasspecificvariablesrelatedtothehotelcompanies suchasthelocation,theinternationalization,theprimaryactivity ofthefirm,etc.,whichallowsustoprovideamorecompletepicture ofthedeterminantsofearningsqualityinthehotelindustry.
Thepaperisstructuredasfollows.Thenextsectionprovides a reviewof thepreviousliteratureand thedevelopmentof our hypotheses;Section3presentstheselectionofthedata,thedef- inition and measurement of earnings quality, the explanatory variablesand,finally,theresearchmethod.Section4presentsthe results.Thelastsectionprovidesourconcludingremarks.
2. Literatureandhypothesis
A majority of companies in the globaleconomy are private andfamily-controlledfirms(LaPortaetal.,1999).InSpain,small andmedium-sizedenterprises(SMEs)represent99%ofbusinesses, employ76.3%ofworkersandcreate65%ofvalueadded(EC,2012).
Spainischaracterizedbytheconcentrationofownership inthe hands of a few large stakeholders.According toLaPorta et al.
(1999),inSpain,thethreelargestshareholdersown51%oftotal shares,while in theUS and theUK this proportion is20% and 19%,respectively.Bankshavetraditionallyalargepresenceascon- trollingshareholdersinfirms.Thetourismindustrysharesthese characteristics.Smallcompaniesarethetypicalbusiness(onlytwo companiesarequoted)andaccesstofinancialresourcesismainly determinedbybanks.
Market participants desire high-quality financial reporting because it reduces information asymmetries, increases overall transparency,andimprovescontracting(WattsandZimmerman, 1986).High-qualityfinancialreportingisapproximatedbyearn- ingsqualitybecauseearningsarethebasisfortheconstructsand measuresusedbyinvestors(Francisetal.,2004).
Mostpreviousstudiesofearningsqualityfocusonlistedcom- panies and developed countries. Only recently research at the intersectionoffinancialreportingqualityandSMEhasattractedthe attentionoffamilyfirmscholars(e.g.,BallandShivakumar,2005;
Burgstahleretal.,2006;VanTendelooandVanstralaen,2008;Katz, 2009;Givolyetal.,2010;Hopeetal.,2013).Thesocialroleplayed bySMEandfamilybusinessesisimportantforthedevelopmentof economies,foremergingcountries,and,especially,fortheSpanish economy.
Theearningsqualityliteraturehasfocusedonthedeterminants ofearningsqualityanditsconsequences(seeDechowetal.,2010 forareview).Recently,empiricalresearchconsideringthequality offinancialreportinginSMEcomparedtolargecompanies(e.g., BallandShivakumar,2005;Burgstahleretal.,2006;Givolyetal., 2010;Hopeetal.,2013)andtheincentivestoimproveearnings qualityhasgainedmomentum(seeAlietal.,2007;Cascinoetal., 2010).Theliteraturedocumentsthat,ingeneral,privatecompanies experiencelowerearningsqualitycomparedtopubliccompanies.
Plausibleexplanationsforlowerearningsqualityincludethelower informationaldemandbystakeholderscomparedtolargercompa- nies(seeBallandShivakumar,2005;Burgstahleretal.,2006;Hope etal.,2013).Theownershipstructure,relationshipsbetweenthe controllinggroupsandotherstakeholders,andconcentrationown- ershipsareextensivelyanalyzedvariablesthatinfluenceearnings quality(see,e.g.,Cascinoetal.,2010).
Twocompetingtheoriesareoftenproposedtoexplainthequal- ityofaccountinginformationinthecaseofSME(see,e.g.Jensen
andMeckling,1976;SalvatoandMoores,2010;Yang,2010;Hope etal.,2013).Accordingtothealignmenthypothesis,greaterowner- shipconcentrationisbeneficialbecauseitreducessevereagency conflictsbetweenownersandmanagers.Theoppositetheory,the entrenchmenthypothesis,predictsthatfamilyownershipisnega- tivelyrelatedtoearningsqualitybecauseconcentratedownership increasestheriskofwealthexpropriationbycontrollingowners attheexpenseofminorityshareholders(seeJensenandMeckling, 1976;SalvatoandMoores,2010,amongothers).
Interestingly,Chenetal.(2012)arguethatthealignmenthypoth- esisand the entrenchmenthypothesis coexist usinga sample of publiclytradedhotelsinTaiwan.Theirevidencesuggeststhatwhen insidermanagerial shareholdings increase, theconvergence-of- interesteffecthasagreaterimpactthantheentrenchmenteffect becauseofanimprovementinfirmperformance(measuredbyROA, ROE,andTobin’sQ)andadecreaseinagencyconflicts.However,the aboveeffectsreverseafteraparticularlevel(i.e.,anoptimalpoint).
Theevidencesuggeststhatwheninsidermanagerialshareholdings increasebeyondtheoptimalpoint,hotelfirmsperformance(mea- suredbyROA,ROE,andTobin’sQ)decreaseconsistentwiththe entrenchmenteffect.
Theliteraturehasalsopaidspecialattentiontothemonitoring roleofboards(andCEOs)ininfluencingonfinancialreportsand thereforeearningsquality.TheCEOiskeyfigureintheproduction andsupervisionoffinancialinformation.Usingsamplesfromthe restaurantindustriesKimandGu(2005),Barberetal.(2006),and Guilletetal.(2012)suggestthatthecompensationofCEOisasso- ciatedwiththefinancialperformanceofthefirm.Motivatedbythe importanceofearningsfigure,TurnerandGuilding(2011)exam- inetheincentivesandmotivationsofhotelownersandoperatorsto engageinearningsmanagementusingaspecificaccountingitem, thechoiceofcapitalizationorexpensingofassetrelatedexpendi- tures.ParteEstebanandSuchDevesa(2011a,b)arguethathotel managersarenotindifferenttoearningsfigures,andtheyprefer toreportsmallprofitsversussmalllosses.Moreover,theydetect differencesinfundamentalvariableswhencomparingfirmsthat reportlossesandprofits.
Cross-countrydifferences in earningsquality maybedue to theeffectsofnationallegalsystems,investorprotections,politi- calinstitutionsandenforcementsystems(e.g.,Leuzetal.,2003;
Burgstahleret al.,2006;Kousenidisetal., 2013).Thequality of earningsinspecific industrieshave alsobeenexaminedin pre- viousresearches.The mostevidence comesfrombank industry (Beattyet al., 2002; Kanagaretnam et al.,2014)and regulatory industries due tomanagers have incentives toengagein earn- ingsmanagementtoobviateriskofinvestigationbygovernmental agencies(Jones,1991; Key, 1997;Hughes etal., 2012).We can alsofindsomestudiesinnonprofitorganizations(LeoneandVan Horn,2005;Eldenburgetal.,2011)andhotelindustry(Turnerand Guilding,2011;ParteEstebanandSuchDevesa,2011a,b).Together, thesestudiesunderlinetheimportanceofnational institutional structuresandindustrycharacteristicsinshapingandsustaining transparentreporting.
Basedontheaboveargument,ourstudyfocusesonasinglemar- ket–Spain–andsinglesector–tourism–whereunlistedfirms playavitalroleintheeconomy.Thetourismindustryisamong themostdynamicindustriesinSpain.Theindustrycontributesto theSpanisheconomy,particularlytoforeignexchangeearnings, employment,andGDP.Thegoalofthispaperistoexaminethe influenceoffirmcharacteristicsonearningsqualityinprivatefirms.
Inthefollowingparagraphs,weexplainourhypothesis.
2.1. Hypothesis
The current financial crisis represents an important factor exogenoustomostfirms.Thefindingsofempiricalresearchonthe
impactoffinancialcrisesonearningsqualityareunclear.Forexam- ple,CallaoandJarne(2011)demonstratedthatearnings-increasing manipulationishigherduring thecrisisforcompanies listedon theSpanishstockmarket. Kousenidisetal. (2013)documented thatearningsquality,ingeneral,hasimprovedduringthecrisis periodusingasampleoflistedEuropeancompanies.However,in thepresenceofincentivestomanageearnings,earningsquality deterioratesratherthanimproves.Inaddition,CohenandZarowin (2007)andStrobl(2013)proposethatincentivesforearningsman- agementinbadtimesarelowerand,therefore,earningsqualityis higher.
IatridisandDimitras(2013)analyzetheeffectoftheeconomic crisisusinga sample oflistedcompaniesthat areauditedby a Big 4 accounting firm in five European countries. Their results revealdifferencesacrosscountrieswithrespecttothedeteriora- tionofearningsquality andvaluerelevance.Portugal,Italy and Greecedisplaystrongertendenciestowardlowearningsquality, and Irelandexhibits evidenceof higher earnings quality.Inter- estingly, the findings for Spain are conflicting. Trombetta and Imperatore (2014) study US listed companiesduring the years 1996–2011fordocumentingthatfinancialcrisessignificantlyaffect earningsmanagementactivitiesoverandabovethebusinesscycle and thedirectionof theeffect varies basedonthe intensityof thefinancialdistress.Therefore,earningsmanagementdecreases whentheintensityofthecrisisislow,whileitincreaseswhenthe crisisisacute.
Thehospitality industryis verysensitivetoeconomic cycles becausecustomersneeddiscretionaryincomestoconsumethese productsand services.Recent empirical studies document that certainfirm characteristicsrespondbettertoadverse economic situations. For example, Chaston (2012),using a sample of 500 small ruralhotels in the UKwith 10–49employees, finds that family-ownedhotelsoutperformednon-familybusinessesduring thecurrentfinancialcrisis.Alonso-AlmeidaandBremser(2013) findthat theeffects ofthefinancialcrisisonhotel performance dependonthefirmmanagementstrategy.Usingasampleofhotels inMadrid,theirresultsdemonstratethathotelsthatfocusonhigh quality,brandimageandaloyalcustomerbasewerebestequipped tohandlethecrisis.Thisstudyexplorestheeffectofthefinancial crisisonearningsquality,andweexpectthatthefinancialcrisis (CRISIS)negativelyinfluencesearningsquality.
H1. Thefinancialcrisisaffectshotels´ıearningsquality.
Accountingandfinanceliteraturesuggeststhatcountry-specific factorsandfirm-specificfactorsexplainaccountingquality.LaPorta etal.(2000)notethatfirms´ıaccountingpracticesoftendepend on the location (geographic and regulatory) of the firm. Leuz etal.(2003)notethatchangesinafirm’sexternalenvironment influencestheexistingrelationsbetweenfirmcharacteristicsand accountingdecision-making. Moreover,although firmsare sub- jectedtothesameregulations,thedifferentconditionsinwhich firmsoperatecanaffectfirmstrategy(accountingpolicychoices and earningsquality).Salvatoand Moores(2010) considerthat accounting quality include a company business and regulatory environment.
Inthehospitality field, severalresearcheshavefocus onthe hotel’scorporatestrategytoengageinasuperiorperformancein ordertoassistinthedecision-makingprocess.Theowner’sstrate- gicdecisionsregardingthelocation(ÁlvarezGiletal.,2001;O’Neill andMattila,2006;Xiaoetal.,2012;Yangetal.,2014),thebrand (Xiaoetal.,2012),thechoicebetweenoperatingunderhotelchain, independentlyorfranchised(ÁlvarezGiletal.,2001;Perrigotetal., 2009),theinternationalization (Zahraet al.,2000;Golovko and Valentini,2011), thehotel age(Álvarez Gil etal., 2001; O’Neill andMattila,2006;Mattilaetal.,2009;Xiaoetal.,2012),andenvi- ronmentpractices(ÁlvarezGiletal.,2001)influenceonfinancial
performance.Thispaperusesasetofvariablesrelatedtothehotel’s corporatestrategythatpotentiallyaffectsearningsquality.Specif- ically, we examine the location, internalization, code and legal constitutionasweexplaininthenextparagraphs.Wecalledthese variables‘businessstrategy’asdependonthehotelierdecision- making.
Thehospitalityliteraturesuggeststhatlocationisacrucialfac- torinthesuccessorfailureofthebusinessinhotelmanagement (seeYangetal.,2014,forareviewinthefield).Thelong-termfixed investmentsthathotelsrequiretooperateplayanimportantrole incasesofchangesinlocation.Accordingtothebusinessorien- tation,hotelsfirmsareclassifiedasurbanhotelsorleisurehotels (O’NeillandMattila,2006).Theformerincludebusinesshotelsthat primarilyofferaccommodations,whilethelatermainlytargetsthe leisuremarket.Inaclosestudy,Mattilaetal.(2009)categorized hotelintourban,suburban,interstatehighway,airport,resort,and smalltowntoexaminethehotelperformance.Theresultsshowdif- ferencesinfinancialperformancebetweenurbanandresortareas oneithercoast.Buildingonpreviousresearch,weclassifyhotels accordingtotheirgeographicallocation:hotelslocatedincoastal areasandhotelslocatedinurbanareas(LOC)toexaminewhether theyreportdifferentlevelsofearningsquality.
Thehospitality industry involvesfirmsthat provideservices suchasaccommodation,restaurants,themeparks,transportation, etc.Theinvestmentrequiredtooperateaccommodationsishigher thanothertourismbusinessessuchasrestaurants,eventplanning, etc.Theseasonalityofthehospitalityindustryrequires firmsto maintainsubstantialinvestmentsininfrastructure. Chen(2010) explainsthatthehighportionoffixedcostsassociatedwithhotel firmsdramaticallyincreasesthesensitivityofthesefirmstobusi- nessconditionsandeconomicdownturns.Theprimaryactivityof thehotelfirm isconsideredwhenreportingthefinancialstate- mentsbecauseitcouldinfluenceonthelevelofaccountingquality.
We distinguishbetween firmsthat declare accommodationsas theirprimaryactivityandthosethatdeclareaccommodationsas theirsecondaryactivity(CODE).
Large hotel chains are constantly expanding their business throughmergers,franchisingormanagementcontracts.TheSpan- ishhotelindustryhasengagedinanexpansivestrategyoverthelast decade.Inthissense,mostpreviousresearchsuggeststhatinterna- tionalizationhaspositiveeffectsontheperformanceofSMEs(Zahra etal.,2000).Thepositiverelationshipbetweenthedegreeofinter- nalizationandtheperformanceofthefirmisduetotheirimproved abilitytoacquirenewanduniqueknowledgeandexperience(Zahra et al.,2000; Golovko and Valentini, 2011), or lower risksfrom diversifyingtheirsourcesofrevenue(Qian,1996),amongothers.
However,GravesandShan(2014)demonstrateanegativeassoci- ationbetweenthedegreeofinternationalizationandROAintheir analysisofAustralianSMEs.Weintroducetheinternationalization (INTER)ofhotelsasanexplanatoryvariableforearningsquality.We expectthatinternationalizedhotelsexhibithigherearningsquality thannon-internationalizedhotels.
Weexaminethebusinessstrategyofthehotelwhentheyset asa company.Thelegalformof anorganization affectsagency relationships,taxes,andcanaffectthebusinessstrategyandthe performance(ScherrandHulburt,2001).Themostcommonlegal formsintheSpanishmarketarePublicLimitedCompanies(PLC) andLimitedLiability Companies(LLC).Bothtypesofcompanies are subject to specific requirements for constitution that then affectthedevelopmentoftheiractivities(suchasaccesstofinan- cial resources,control over financial information, requirements ofinformation disclosure,etc.).Therefore, thelegal constitution (LEGAL)caninfluenceearningsquality.
Ournexthypothesisinvestigateswhetherbusinessstrategy(the primaryactivity,thegeographicallocation,theinternationaliza- tion,andthelegalconstitution)isassociatedwiththequalityof
reportedearnings.Hence,oursecondhypothesisisformulatedas follows:
H2. Businessstrategyaffectshotels´ıearningsquality.
Theeffectsofmanagerialstyle,ownershipstructure,ownership concentration,theCEOasamajorshareholder,andtheindepen- denceoftheboardonearningsqualityreceivedspecialattention within theearnings quality and family firm literatures. Agency theory predicts two possible effects related to ownership con- centration.First,lessseparationbetweenownershipandcontrol leadstoless manipulationofearningsforopportunistic reasons and,therefore,tohigherqualityearningsforecasts.Second,greater ownership concentration can result in executive entrenchment (see,e.g.,SalvatoandMoores,2010forareview).
Theevidencelinkingownershipstructuretoearningsqualityis mixed.Somestudiesshowthatfamilyfirmstendtoexhibithigher earningsqualitythannon-familyfirmsandexplainthisfindingby arguingthatfamilyownershipreducesagencyconflictsbetween ownersandmanagers(e.g.,Wang,2006;Alietal.,2007).Forexam- ple,Wang(2006)findsthatearningsqualityishigherforfamily firmsthanfornon-familyfirmsintheU.S.Alietal.(2007)useasam- pleofU.S.familyfirmstodemonstratethathigherqualityfinancial disclosuresarefollowedbymoreanalystswhothen tradetheir stockswithnarrowerbid-askspreads.However,existingstudies suggestthatprivatefirmsexhibitlowerearningsqualitythanpub- licfirmsbecauseoftheentrenchmenteffect.Forexample,Balland Shivakumar(2005)arguethatearningsqualityislowerinprivate companiesthaninpubliccompaniesbecauseofdifferentmarket demands.Hopeetal.(2013)demonstratethat,onaverage,pub- licfirmsexhibitsignificantlyhigherearningsqualitythanprivate firmsbecauseinvestorsandstakeholdersinpublicfirmsdemand higherqualityfinancialinformation.Givolyetal.(2010)alsofind differencesinearningsqualitybetweentwotypesfirms:publicly tradedequityandprivatelyheldequity.Theevidencesuggeststhat privateequityfirmshavelowerpropensitytomanageincomethan publicequityfirms.
Xiaoetal.(2012)focusontheinfluenceofhotelownersonthe level offinancialperformance. Theydemonstratethatthehotel owner’sexpertiseinimplementingsuperiorstrategiesregarding marketsegment,brand,operation,andlocation(i.e.,thestate)are criticaltohotelfinancialperformance(revenuesandprofits).Brady andConlin(2004)comparedtherevenueperformanceoftwodif- ferenttypesofhotelowners:realestateinvestmenttrusts(REITs) andnon-REITs.ThefindingssuggestthatREIT-ownedhotelsper- formbetter,onaverage,thannon-REITproperties.Perrigotetal.
(2009) compared the efficiency of pluralized chains (chains in franchisedunits)withthatofcompany-ownedunits.Theresults indicatethatpluralizedchainsreportsignificantlyhigheraverage efficiencyscoresthanpredominantlycompany-ownedhotels.
Ournexthypothesisinvestigateswhethertheownershipstruc- tureandconcentrationofownershipareassociatedwiththequality ofreportedearnings.Weexpectmanagement styletoinfluence earningsqualitybutwedonotpredictadirectionoftherelation- ship.
H3. Ownershipstructureaffectshotels´ıearningsquality.
Agencytheorysuggeststhatauditingservicesserveasamon- itoringmechanismtoreducethecoststhatarisefromtheconflict ofinterest betweenprincipalsand agents.Auditing servicesare usuallyviewedasamonitoringmechanismthatfirmsdemandto provideevidencethattheyproducereliablefinancialstatements forstakeholders(mainlyinvestors,creditors,etc.).Todate,theevi- dencegenerallysupportstheaboveargument.Existingresearch has focused on the effect of audited firms versus non-audited firmsonearningsqualityandonwhetheraBig4auditorprovides
superiorauditqualitythananon-Big4firm(DeAngelo,1981;Van TendelooandVanstralaen,2008;Cascinoetal.,2010).
For example, Cascino et al. (2010) using a sample of listed familyfirms, showthat thepresenceof a renownedaudit firm increasesearningsquality.Thatis,theargumentsoftheseminal paperbyDeAngelo(1981)aretestedbecausehigherauditqual- ityenhancesa higher degreeof compliance. VanTendeloo and Vanstralaen(2008)examinea sampleofprivateEuropeancom- panies(Belgium,Finland,France,Netherlands,SpainandtheUK) andfindthatlargeauditingfirms(theBigauditors)constrainearn- ingsmanagementpracticestoagreaterextentincountrieswith ahightaxalignment(Belgium,Finland,FranceandSpain).They explaintheseresultsbyanalyzinglitigationrisk.Evenincountries withastronginvestorprotection,investors,financialanalystsand regulatingauthoritiesdonotscrutinizeprivatecompanies;there- fore,theirlitigationriskislowerthanforlistedcompanies.Kosiand Valentincic(2013)analyzeprivatefirmsusingfinancialreporting fortaxsavingsinsteadofforcommunicatingobjectivesandtheir resultshighlightthatauditorsconstrainfirmsfromobtainingtax benefits.Burgstahler etal.(2006)andFrancisandWang(2008) studytheeffectofauditquality,amongothervariables,inearnings qualityincross-countryanalysis.FrancisandWang(2008)findthat Big4auditorsenforcehighqualityearningsonlyincountrieswith stricterinvestorprotectionregimes.
Our nexthypothesis investigates whether audits are associ- atedwiththequalityofreportedearnings.First,weconsiderthat auditedcompanies(AUD)reporthigherearningsqualityand,sec- ond,wetestwhetherthesizeoftheauditingfirm(BIG4)ispositively associatedwithearningsquality.
H4. Auditcharacteristicsaffecthotels´ıearningsquality.
Finally, Control Variables have been included to reduce the noiseinthemeasurementoftheimpactofaccountingchangeson earningsqualityincludingsize,leverage,losses,deviationofsales revenuesanddeviationofcashflow(seeFrancisetal.,2004;Gaio, 2010).
Firmsizeisoftenselectedasacontrolvariableinresearch.On theonehand,firmsizeisrelatedtooperationefficiency.Onthe otherhand,firmsizeiscorrelatedwiththelevelofcashflowand accruals,whichareinherentlylinkedtoearningsquality.Thecoef- ficientofthevariableSIZEisexpectedtobepositive(Francisetal., 2004;Gaio,2010).
Leverageisanothercommoncontrolvariable.Leveragerepre- sentsthetrade-offbetweentaxbenefitsandbankruptcycosts.In particular,thelevelofleveragereflectsthefirm’spotentialriskand influencesthefirm’sreportingandaccrualaccountingpolicies.The coefficientofthevariableLEVisexpectedtobepositive(Francis etal.,2004;Gaio,2010).
Finally,weintroducetheeffectoflossesinearningsthattyp- ically reflect economic distress in a business. We consider the standarddeviationofthesalesrevenuesd(SALES)andthestandard deviationofthecashflowd(CFO)(Francisetal.,2004).Weexpect theincidenceofLOSS,d(SALES),andd(CFO)tobenegativelyassoci- atedwithaccountingquality(Francisetal.,2004;Gaio,2010).
3. Data,variablesandmethodology 3.1. Selectionofthesample
The study draws on a comprehensive database (SABI) that includesmostSpanishfirmsandisadequatetoexaminesmallbusi- ness.Thedatabaseincludesfinancialstatementsand qualitative variablesrelatedtoownershipstructure,auditvariables,andfirm specificfactors.Oursampleincludesallhotelfirmsreportingfinan- cialinformationannuallyovertheperiod2000–2011.Weselect companiesdeclaringaccommodationastheirprimaryactivityas
wellastheirsecondaryactivity.Wedeletedobservationwithtotal assetslessthan1.Wealsodeletedobservationswithmissingval- uesonearningsandtotal assets(i.e.,dataerrors).Theselection proceduresresultsinasampleof1805firmsand21,660firm-year observations.Tomitigatetheeffectsofoutliersinourregressions, earningsandoperatingcashflowsarewinsorizedat5%.
ThesampleisformedmainlybySMEfirmsbecauseisthepre- dominantfirmsizeofSpanisheconomy.Thecharacteristicsofthe sampleareasfollows.Mainlyfamilyfirms(around55%)compared tootherownershipstructuressuchasmutualfunds,institutional investors,etc., constitutestheownershipstructure. Mostofthe hotelinthesampleisnotinternationalizedduetotheirsmallsize.
Itisalsonotedthatmostofthefinancialstatementsarenotaudited (morethan80%)becausetherequirementtobeauditedislinked directlywiththesizeofthefirm.
3.2. Measurementofearningsquality
Earnings quality is one of the most important accounting researchtopicofthelastfewdecades.However,despiteitsrel- evance,thereisneitheranagreed-uponmeaningoftheconcept nora generallyacceptedapproach tomeasuringearningsqual- ity(SchipperandVincent,2003).Earningsqualityisconsidereda multidimensionalconceptthatisdifficulttomeasure,andrecent empirical researchevaluates it by consideringvarious earnings attributes(Francisetal.,2004;Gaio,2010;Dechowetal.,2010;
Kousenidisetal.,2013).
Wemeasureearningsqualityusingaccounting-basedattributes thatdonotdependonmarketperceptions(e.g.,Leuzetal.,2003;
Francisetal.,2004;Burgstahleretal.,2006;Gaio,2010;Kousenidis etal.,2013)becauseoursampleincludesmostlyprivatefirms(only twofirmsarelisted).
Amongtheaccounting-basedattributes,time-seriesproperties of earnings indicate how profits are distributed over time and thestatisticalcharacteristicsoftheprocessthatgeneratesearn- ings(Schipper andVincent, 2003).Specifically,weconsiderthe followingearnings attributes:thepersistenceof earnings,which capturestheextenttowhichagiveninnovationproducesfuture earnings;thepredictiveabilityofearnings,whichisafunctionof thedistributionoftheinnovationseries;thevariabilityofearnings, whichmeasuresthetime-seriesvarianceofinnovationsdirectly.
Additionally,weincludeearningssmoothingwhichmeasuresthe intentionalattemptsofmanagerstoeliminatefluctuationsinearn- ings.
Earningspersistenceisassociatedwithstability,sustainability, and recurrence of earnings over time. This attribute could be definedasthesystematicbehaviorofearnings,andpersistentearn- ingsareviewedasdesirablebecausetheyarerecurring(see,e.g., Francisetal.,2004).
Itisdefinedastheslopecoefficientestimatedfromafirstorder autoregressivemodel(AR(1))forannualearnings.
Xj,t=0,j+1,jXj,t−1+vj,t (1)
whereXj,tandXj,t−1arefirmj’searningsinyeartandt−1,respec- tively,andcoefficient1,jcapturesfirmj’spersistenceofearnings.
Valuesof1,jclosetooneimplyhighlypersistentearnings,while valuesof1,jclosetozeroimplyhighlytransitoryearnings.Persis- tentearningsareviewedashigher-qualityearnings.
Thismeasure,basedonLipe(1990),hasbeenutilizedbymost researchersstudyingearningspersistence(e.g.,Francisetal.,2004;
DichevandTang,2009;Cascinoetal.,2010;Gaio,2010;Kousenidis etal.,2013).
Earningspredictabilitymeasuresthe abilityof earningstobe predicted.Thisattributeismeasuredbasedonthevarianceofthe
earningsshocks,wherehighervarianceimplieslowerpredictabil- ity.WeusethesquarerootoftheerrorvariancefromEq.(1)
Predictability=
2
ˆvj,t
Large (small) values of predictability imply less (more) predictableearnings.Earningsthataremorepredictablearecon- sideredhigherqualityearnings.
Earningspredictabilityhasbeenmeasuredwiththis method, proposedinitially byLipe (1990),by many researcherssuchas Francis et al. (2004), Cascino et al. (2010), Gaio (2010), and Kousenidisetal.(2013),amongothers.
Earningsvariability ismeasuredas thestandard deviationof earnings.
Variability=(Xj,t)
whereXj,tisfirmj’searningsinyeart.
Higher(lower)valuesrepresenthigher(lower)levelsofearn- ingsvariability,whichareinterpretedaslower(higher)earnings quality.
EarningsvariabilityhasbeenusedbyresearcherssuchasFrancis etal.(2004);FrancisandWang,2008orDichevandTang(2009).
Earningssmoothingisconsideredamanipulativetechniqueto reducenaturalearningsvariability.Inthisview,smootherearnings implylowerearningsquality(see,e.g.,DechowandSkinner,2000;
Zeghaletal.,2012).Itisusuallymeasuredastheratioofearnings variabilitytocashflowvariability.
ESM= (Xj,t) (CFOj,t)
whereXj,tisfirmj’searningsinyeartandCFOj,tisthecashflow fromoperationsinyeart.
Lower(higher)valuesindicatehigher(lower)variabilityincash flowsthaninearningsand,therefore,ahigher(lower)levelofarti- ficialearningssmoothing.
Scholarswhoemploythismeasureofearningssmoothing,pro- posedbyLeuzetal.(2003),includeFrancisetal.(2004),Burgstahler etal.(2006),VanTendelooandVanstralaen(2008),Cascinoetal.
(2010),Gaio(2010),andKousenidisetal.(2013),amongothers.
Table1PanelA,summarizesthedefinitionandtheconstruction ofeachattribute.Followingpreviousresearch(e.g.,Leuzetal.,2003;
Gaio,2010),wecomputeanaggregateearningsqualitymeasure.
Higherrankingssuggesthigherlevelsofearningsquality;therefore, highervaluesoftheaggregatemeasureofearningsquality(EQ) suggesthigherearningsquality.
3.3. Measurementofthedeterminantsofearningsquality
Asweexplainedbefore,weexaminevariablesthatarepoten- tially associated with earnings quality and sort these into five categories:marketvariables,businessstrategy,ownershipstruc- ture,auditfunctionandcontrolvariable.Inthenextparagraph,we explainthedefinitionofthevariables.
First,weexploretheeffectofthefinancialcrisisonearnings quality(CRISIS).Theeconomic crisisbegan in 2007inthebank system.Themostcommonapproachtoanalyzetheeffectofthe financialcrisesintheliteratureisthroughbinaryvariablesbased onunderlyingfinancialvariable.Buildingonthispreviouslitera- ture,wecreateadummyvariablethattakesthevalueof1forthe crisisperiod2008–2010and0forthepre-crisisperiod.Then,we considerthatthe2008isthefirstyearbeingaffectedbythecrisis (seeCallaoandJarne,2011;Chaston,2012;Alonso-Almeidaand Bremser,2013).Inasensitivityanalysis,wedatethefinancialcri- sisto2009insteadof2008becausetherealeffectsofthefinancial
crisisonthehotelindustrycouldbeginineither2008or2009.We obtainsimilarresults.
Thelocationvariable(LOC)ismeasuredaccordingtothegeo- graphical location of the hotel: coastal areas and urban areas (O’NeillandMattila,2006;Mattilaetal.,2009).Toidentifythepri- maryactivityofthehotel(CODE),weusethemainactivitydeclared bythehotelwhenreportingfinancialaccountingstatements.That is,weuseadummyvariablethattakesthevalueof1ifthehotelfirm declaresaccommodationsastheirprimaryactivityandthevalueof 0isthehoteldeclaresaccommodationsastheirsecondaryactivity.
Thelegalconstitution(LEGAL)allowustodiscriminatebetweenthe legaladoptedbythefirmwhenestablishingasacompanyinthe market.Thisvariableisadummyvariablethattakesthevalueof1 iflegalofthefirmisPLC,or0otherwise.
Perrigot et al. (2009) and Xiao et al. (2012) provide useful reviewsofthehospitalityliteratureaboutthedefinitionofowner- shipstructure.Thedatabaseusedinthispaperlimitstheavailable ownershipstructuredataaspreviousstudies.Forexample,Chen et al. (2012)focus on only7 tourist hotel to studythe impact of insidermanagerial ownershiponcorporateperformance and Perrigotetal.(2009)focuson15hotels(6ofthemarepresented apluralformand5ofthemarepredominantlycompany-owned units)toanalyzetheirefficiency.Theexemptionisprovidedby Xiaoetal.(2012)thatuse2012hotelsacrosstheUSAfortheperiod 2003–2005.
Inourstudy,wedifferentiatebetweenpureformssuchasfamily ownership(FAM)andotherformssuchasinstitutionalownership andmixedformsthatcombinealternativestructures(familyand financial,familyandindustrial,etc.).Additionally,weconsidera variable measuringownership concentration.In general,family firmsaremoreconcentratedthannon-familyfirms.Thevariable CONCrepresentsshareholderownershipofmorethan25%equity directly(first level)or indirectly(see, e.g., Salvatoand Moores, 2010).
Thevariablesrelatedtotheauditfunctionaremeasuredfollow- ingtheaccountingliterature(DeAngelo,1981;Burgstahleretal., 2006;VanTendelooandVanstralaen,2008;Katz,2009;Givolyetal., 2010;Cascinoetal.,2010;KosiandValentincic,2013).Thatis,we defineAUDvariableasadummyvariablethattakesthevalue1for auditedfirmsand0fornonauditedfirm.ThevariableBIG4takes thevalue1forfirmsauditedbyBig4companiesand0forfirmsnon auditedbyBig4companies.
Webasedonpreviousearningsqualityworkstodefineandmea- surethecontrolvariables(see,e.g.Francisetal.,2004;Burgstahler etal.,2006;Cascinoetal.,2010;Gaio,2010).Thevariablesize(SIZE) ismeasuredbythelogarithmoftotalassets(see,e.g.Francisetal., 2004;Cascinoetal.,2010;Gaio,2010).ThevariableLeverage(LEV) ismeasuredbytheratiooftotalliabilitytototalassets(see,e.g.
Francisetal.,2004;Cascinoetal.,2010;Gaio,2010).Thedefinition ofloss(LOSS)isincludedasthenumberofyearsoflossesrelative tothetotalnumberofyearsofoperation(Francisetal.,2004;Gaio, 2010).Finally,thestandarddeviationofthesalesrevenuesd(SALES) andthestandarddeviationofthecashflowd(CFO)areincludedin themodels(seee.g.Francisetal.,2004;Gaio,2010).Mostofthese variableshavebeenusedtoexaminethefactorsthatenhancethe hotelperformance.Forexample,Kimetal.(2012)showthatfirm size,debtleverageandgrowtharesignificantfactorsthataffectthe systematicriskofahotelfirm.
Table1PanelB,summarizesthedefinitionandtheconstruction ofeachvariableemployedinthesecondstage.
3.4. Methodology
Wecomputeearningsqualitymeasuresoverfirm-specificfive- year windows, from t−4 to t, following Francis et al. (2003).
Therefore, we obtain earnings quality measures for the period
Table1
Variablesdefinitions.
PanelA:Earningsqualitymeasures
Earningsattributes Definition Calculation
Earningspersistence Stabilityofearnings(degreetowhichunexpectedearnings fromoneperiodpersistinfutureearnings)
Firstorderautoregressivemodel(AR(1))forannual earnings.Xj,t=0,j+1,jXj,t−1+vj,t
Earningspredictability Abilityofearningstobepredicted SquarerootoftheerrorvarianceofAR(1)
2(ˆvj,t) Earningsvariability Realvolatilityofearnings Standarddeviationofearnings(Xj,t)
Earningssmoothing Intentionalreductioninearningsvariability Standarddeviationofearningsdividedbythestandard deviationofcashflowsESM=(CFO(Xj,t)
j,t)
PanelB:Explanatoryfactors Marketvariables
CRISIS Financialcrisis.Dummyvariabletakingthevalueof1for(2008–2010)period,or0otherwise Businessstrategy
CODE Dummyvariabletakingthevalue1forfirmsdeclaringaccommodationastheprimaryactivity,or0otherwise LOC Dummyvariabletakingthevalue1forfirmslocatedincoastareas,or0otherwise
INT Dummyvariabletakingthevalue1forinternationalizedfirms,or0otherwise LEGAL Dummyvariabletakingthevalue1iflegalofthefirmisPLC,or0otherwise Ownerstructure
FAM Dummyvariabletakingthevalue1forfamilyfirms,or0otherwise
CONC Dummyvariabletakingthevalue1forfirmsthatshareholdershavemorethan25%equity Auditcharacteristic
AUD Dummyvariabletakingthevalue1forauditedfirms,or0otherwise BIG4 Dummyvariabletakingthevalue1forfirmsauditedbyBig4,or0otherwise Controlfactors
SIZE Logoftotalassets
LEV Financialleverage,measuredastotalliabilitiesdividedbytotalequity
LOSS Numberofyearswithnegativenetincomebeforeextraordinaryitemsdividedbythetotalnumberofyearsforeachfirm d(SALES) Salesvolatility.Salesvolatilitymeasureasthestandarddeviationofsalesrevenuesscaledbytotalassets
d(CFO) Cashflowvolatility.StandarddeviationofCFOscaledbytotalassets
2004–2011foreachcompany.Tocalculatetheaggregateearnings qualitymeasure,werankfirmsonascalefrom0to100according toeachoftheearningsattributes,andwecomputetheEQvariable byaveragingitsindividualearningsqualitymeasures.
Toexploretheearningsqualitydeterminants,webeginwith aunivariateanalysis.First,we calculatethePearsonand Spear- mancorrelationstoobservetherelationshipsbetweentheearnings qualitymeasureandtheexplanatoryvariables.Second,thet-mean testandthenon-parametricU-MannWhitneytestdetectsstatisti- callysignificantdifferencesinthemeanandmedianoftheearnings qualitymeasuredrivenbytheindependentvariables.Theytestthe nullhypothesisthattwopopulationssharethesamedistribution, thatis,whethertherearedifferencesinonevariable(earningsqual- ity)betweentwogroupsthatareindependent(splitonthebasisof theexplanatoryvariablesinthisstudy).
Inadditiontotheunivariateanalysis,weestimateapaneldata regressionusingOLS.Allregressionsincludeyeardummyvariables andstandarderrorsareclusteredbyfirmtocorrectforserialcor- relation.Tomaximizeobservationnumberthroughtheregression analysis,weincludethevariablesblockbyblock.Forexample,the dataforauditorsize(BIG4)arenotavailablefortheentiresample becausemostfirmsare notaudited firms,and alsotheowner- shipstructurevariablereducesthesamplesize.Thenextsection presentsourmainresults.
4. Results
4.1. Descriptivestatistics
Table2presentsthedescriptivestatisticsforthemainvariables.
PanelAreportstheattributesincludedinthedependentvariable, theindexof earningsquality. Persistencehasa mean (median) valueof0.08(0.04),Predictabilityhasamean(median)valueof 0.03(0.024),Smoothnesshasamean(median)valueof0.59(0.59),
andVariabilityreportamean(median)valueof0.05(0.03).Panel Bliststhedependentvariable(EQ)andthecontrolvariables.The descriptivestatisticsforthemainexplanatoryvariablesrelatingto earningsqualityindex(EQ),aredisplayedinTable4.
4.2. Univariateanalysis
Table3reportsthePearsonandSpearmancorrelationmatrices fortheearningsqualityindex,theexplanatoryandthecontrolvari- ables.Theearningsqualityindexispositivelyassociatedwithfirms declaring accommodation astheprimary activity (CODE), firms locatedincoastalareas(LOC),internationalizedfirms(INT),firms establishedasPLCs(LEGAL),auditedfirms(AUD)andthoseaudited byBig4auditors(BIG4).Allthesecoefficientsarestatisticallysig- nificant(p<0.01).
Table2
Descriptivestatistics.
PanelA:Attributesofearningsquality
Persistence Predictability Smoothing Variability EQ
n 14,440 14,440 14,440 14,440 14,440
Mean 0.081 0.039 0.592 0.052 0.498
Median 0.045 0.024 0.592 0.035 0.508
StdDev 0.552 0.045 0.419 0.051 0.205
Q10% −0.628 0.007 0.129 0.011 0.305
Q90% 0.874 0.088 1.000 0.119 0.678
PanelB:Controlfactors
LOSS SIZE LEV d(SALES) d(CFO)
n 14,440 14,440 14,440 14,440 14,440
Mean 0.331 3.305 2.039 0.223 0.117
Median 0.250 3.274 0.819 0.112 0.079
StdDev 0.288 0.676 3.586 0.281 0.111
EQconsiderspersistence,predictability,variabilityandsmoothing.
PleaseseeTable1forvariablesdefinitions.
Table3 Pearson(belowdiagonal)andSpearman(abovediagonal)correlationcoefficientsbetweendependentvariable(earningsquality),explanatoryandcontrolvariables. EQCRISISCODELOCINTLEGALAUDBIG4FAMCONCSIZELEVLOSSd(SALES)d(CFO) EQ–−.002.118***.027***.038***.074***.033***.072***−.015***−.048***.267***−.002−.166***−.433***−.669*** CRISIS−.003–−.012.000.000.000.000.003.000.000.025***−.038***.000−.161***.042*** CODE.124***−.012–.091***−.021**.126***.116***.030−.099***−.015.208***−.067***.035***−.149***−.125*** LOC.029***.000.091***–−.007.091***.144***.127***−.094***.014.205***−.022***−.050***−.121***−.023*** INT.035***.000−.021**−.007–.047***.106***.140***−.098***.046***.103***.033***−.033***−.016*−.011 LEGAL.076***.000.126***.091***.047***–.262***.006−.231***−.067***.356***−.107***−.059***−.122***−.074*** AUD.036***.000.116***.144***.106***.262***–..421***−.002.532***.005.022***−.124***.023*** BIG4.080***.003.030.127***.140***.0060.45***–−.244***.065***.233***.046**.204***.064***.179*** FAM−.015***.000−.099***−.094***−.098***−.231***−.421***−.244***–.038***−.396***−.011−.126***.148***−.038*** CONC−.049***.000***−.015***.014**.046**−.067**−.002*.065***.038***–−.014*.037***.047***.034***.059*** SIZE.277***.025***.198***.198***.112***.329***.565***.245***−.399***−.015–.054***−.016*−.450***−.252*** LEV−.013−.029***−.046***−.011.011−.108***.001.059***.007.024**.046***–.061***.049***.043*** LOSS−.142***.000.041***−.033***−.028***−.040***.039***.221***−.139***.047***−.063***.066***–.000.103*** d(SALES)−.399***−.143***−.183***−.123***−.014*−.159***−.110***.133***.123***.022*−.456***.079***.007–.517*** d(CFO)−.585***.039***−.145***−.034***−.008−.089***.012.203***−.024**.052***−.230***.086***.109***.508***– EQconsiderspersistence,predictability,variabilityandsmoothing. *,**,and***denoteslevelofsignificanceatthe0.10,0.05and0.01levels,respectively. PleaseseeTable1forvariablesdefinitions.
In contrast, the earnings quality index is negatively associ- atedwithfamilyownership(FAM)andownershipconcentration (CONC);thecoefficientstatisticallysignificant(p<0.01).Inaddi- tion,thecoefficientrelatedtoCRISISarenegativelyassociatedwith earningsqualitybutisnotstatisticallysignificant.
Thecontrol variables exhibitmostly significantrelationships withearningsqualityandareconsistentwiththeresultsofprevi- ousresearch.Largerfirms(SIZE)havehigherearningsqualityand firmswithhighersalesvolatility(d(SALES)),greateroperatingcash flowvolatility(d(CFO)),andahigherincidenceoflosses(LOSS)have lowerearningsqualityrankings(Francisetal.,2004;Gaio,2010).
Table3alsorevealsthattheindependentvariablesarenothighly correlated.
Interestingly,theresultsallowustocharacterizefamilyfirms.
The coefficient on FAM is negatively associated with firm size (SIZE)indicatingthat,ingeneral,familyfirmsaresmaller.More- over, the coefficient onFAM is also negatively associated with internationalization (INT), withestablishment as a PLC(LEGAL), whichcharacterizeslargercompanies,withaudits(AUD)andBig4 auditors(BIG4).Incontrast,thecoefficientonFAMispositively associatedwithCONCindicatingthatfamilyfirmsarecharacterized bygreaterconcentration.
Table4 displaysthedifferencesin earningsquality between groupsoffirmsdividedbytheexplanatoryvariablesinTable1.The t-mean andU-MannWhitneytestsareparticularlyusefulwhen assessingdifferencesbetweentwoindependentgroups.
Table4revealsdifferencesinearningsqualityindexaccording to the firm’s business strategy. Specifically, we observe differ- ences between firms declaring accommodation as the primary activity(mean=0.508)andfirmsdeclaringaccommodationasthe secondary activity(mean=0.467), firmslocatedincoastalareas (mean=0.501)and firms locatedin urbanareas (mean=0.492), internationalhotels(mean=0.539)anddomestic(mean=0.498), and PLC firms (mean=0.513) and non PLC firms (mean=0.49).
Thesedifferencesarestatisticallysignificant(p<0.01)accordingto thet-mean.Thenon-parametrictest,U-MannWhitney,alsoreports differencesbetweenthesevariables.
Theevidenceforownershipstructurerequiresattention.Fam- ilyfirms(mean=0.50)exhibitslightlowerearningsqualitythan othertypesofownership(mean=0.504).Thepercentageofequity heldbyshareholders(CONC)alsoproducesstatisticallysignificant differencesintheearningsqualityindex.
Interestingly,firmsthatareaudited(AUD)andfirmsauditedby theBig4(BIG4)exhibithigherEQindexvaluesthanfirmsthatare notauditedorareauditedbysmalleraccountingfirms.Thediffer- encesarestatisticallysignificant(p<0.05).Thisevidencesupports previousstudiesdocumentingthatauditorsizeisdirectlylinkedto auditqualityand,therefore,toearningsquality(DeAngelo,1981;
FrancisandWang,2008).
4.3. Multivariateanalysis
Table5presentstheresultsofourmultivariateanalysis.The dependentvariableinallmodelsistheearningsqualityindex(EQ).
Theexplanatoryvariablesareintroducedbyblockstomaximize thesamplesize(theintroductionoftheenforcementvariablesand ownershipstructuresignificantlyreducesthesamplesize).
Model1(thebasicmodel)includesindicatorsofmarketcon- ditions,businessstrategy,andcontrolvariables.Models2and3 addthevariables ofownership structureand thepercentageof equityheldbystakeholderstothebasicmodel,respectively.Model 4completesthebasicmodelwithauditcharacteristics,andModel 5presentsthefullmodel.
ThecoefficientforCRISISisnegativeandstatisticallysignificant (p<0.01).Inanadditionalanalysis,wedatethefinancialcrisisto 2009insteadof2008becausetherealeffectsofthefinancialcrisis
Table4
t-TestandU-MannWhitneytestforearningsqualitydifferences(EQ).
Variable Type N % Mean Median Stddev Type N % Mean Median Stddev p-value
t-mean U-MannWhitney Marketvariables
CRISIS 2008–2010 7220 0.500 0.498 0.503 0.141 2004–2007 7220 0.500 0.499 0.504 0.139 0.75 0.87 Businessstrategy
CODE primary 11,016 0.763 0.508 0.513 0.137 Second. 3424 0.237 0.467 0.470 0.146 0.00 0.00
LOC Coast 10,352 0.717 0.501 0.506 0.139 Urban 4088 0.283 0.492 0.497 0.143 0.00 0.00
INT Yes 208 0.014 0.539 0.558 0.154 No 14,232 0.986 0.498 0.503 0.140 0.00 0.00
LEGAL PLC 5153 0.357 0.513 0.520 0.135 No 9287 0.643 0.490 0.495 0.142 0.00 0.00
Ownerstructure
FAM Yes 5184 0.548 0.500 0.504 0.140 No 4272 0.452 0.504 0.510 0.139 0.15 0.15
CONC +25% 6832 0.914 0.498 0.504 0.204 −25% 640 0.086 0.522 0.532 0.193 0.00 0.00
Auditcharacteristic
AUD Yes 2464 0.171 0.509 0.513 0.135 No 11,976 0.829 0.496 0.502 0.141 0.00 0.00
BIG4 Yes 721 0.293 0.516 0.518 0.141 No 1743 0.707 0.493 0.493 0.132 0.00 0.00
EQconsiderspersistence,predictability,variabilityandsmoothing.
PleaseseeTable1forvariablesdefinitions.
Table5
Resultsofthepanelregressionsmodels.
Model1 Model2 Model3 Model4 Model5
coef. p-value coef. p-value coef. p-value coef. p-value coef. p-value
C 0.552 0.000 0.539 0.010 0.555 0.000 0.539 0.000 0.543 0.000
CRISIS −0.024 0.000 −0.026 0.000 −0.027 0.000 −0.024 0.000 −0.030 0.002
CODE 0.006 0.007 0.008 0.010 0.005 0.121 0.006 0.005 0.003 0.679
LOC 0.008 0.000 −0.002 0.499 0.002 0.461 0.008 0.000 0.004 0.569
INT 0.019 0.015 0.017 0.037 0.007 0.040 0.020 0.012 0.015 0.020
LEGAL 0.008 0.000 0.090 0.000 0.009 0.000 0.007 0.001 0.009 0.091
FAM −0.090 0.019 −0.019 0.002
CONC −0.007 0.090 −0.007 0.080
AUD 0.020 0.020
BIG4 0.030 0.020 0.040 0.050
SIZE 0.022 0.000 0.023 0.000 0.021 0.000 0.026 0.000 0.019 0.000
LEV 0.000 0.163 0.001 0.122 0.001 0.008 0.000 0.213 0.001 0.141
LOSS −0.043 0.000 −0.028 0.000 −0.022 0.000 −0.043 0.000 −0.011 0.090
d(SALES) −0.062 0.000 −0.072 0.000 −0.075 0.000 −0.059 0.000 −0.134 0.000
d(CFO) −0.620 0.000 −0.619 0.000 −0.619 0.000 −0.616 0.000 −0.557 0.000
Yeardummy Included Included Included Included Included
Obs. 14,440 9456 7472 2464 2224
Prob>F 0.000 0.000 0.000 0.000 0.000
R-squared 0.379 0.379 0.376 0.259 0.323
Thedependentvariableisearningsqualityindex(EQ)thatconsiderspersistence,predictability,variabilityandsmoothing.
Reportedsignificancelevels(*,**,and***for0.10,0.05and0.01levels,respectively).
PleaseseeTable1forvariabledefinitions.
onthehotelindustrycouldbeginineither2008or2009.Overall, theresultsfromthesevarioussensitivitytestsindicatenomajor threatstothemainfindings.
Thecoefficientsofthebusinessstrategyvariablesarestatisti- callysignificantacrossthemodels(p<0.01)andtheassociationis inthesamedirectionastheunivariateanalysis.Ourresultssup- porthypothesisH2inthesensethatthebusinessstrategyaffects earningsquality.
ThecoefficientsforFAMandCONCarenegativeandstatistically significant(p<0.05).Ourresultsprovidesomesupportforhypoth- esisH3inthesensethattheownershipstructureaffectsearnings quality.ThecoefficientsforAUDandBIG4arepositiveandstatisti- callysignificant(p<0.05).OurresultssupporthypothesisH4.That is,auditfunctionaffectsearningsquality.
Finally,thecoefficientsforthecontrolvariablesarestatistically significant(p<0.01).ThereisapositivelyrelationshipbetweenSIZE andearningsquality.Incontrast,therearenegativelyrelationships betweenLOSS,d(SALES)andd(CFO).ThecoefficientofLEVispos- itivelyassociatedwithearningsquality,butitisnotstatistically significantacrossmodels.
Itis notedthat theearningsquality indexis measuredwith fourattributes:persistence,predictability,variabilityandearnings
smoothing. As there is considerable debate among researchers about the best measure of earnings quality, in an additional analysisweredefiningearningsqualityindex(EQ)toexcludeearn- ingssmoothing. The results(not reported) are consistent with Tables4and5.
5. Conclusions
Earningsquality is a fundamental concept for the decision- makingofstakeholders.Inaddition,itisimportantforstandard setterstoprovideusefulinformationfromfirmstoinvestors.For example,thejointIASB/FASBprojectontheconceptualframework (2006)statesthattheobjectiveoffinancialreportingistoprovide usefulinformationtousersmakingresourceallocationdecisions.
Forthisreason,earningsqualityhasreceivedmuchattentioninthe literature.
Inthehospitalityfield,theliteraturehaspointouttheimpor- tance of earnings figure to arrange contracts between related parties(KimandGu,2005;Chen,2010;TurnerandGuilding,2011;
Guilletetal.,2012;Chenetal.,2012;Xiaoetal.,2012).Turnerand Guilding(2011)explainthatthemostcommonwayfordetermin- ingahoteloperator’sremunerationisbasedonahotel’sreported