International Earnings Comparability

Transcription

1 Inernaonal Earnngs Comparably Chrsof Beuselnck Tlburg Unversy CenER and deparmen of Accounancy Phlp Joos Tlburg Unversy CenER and deparmen of Accounancy Sofe Van der Meulen Tlburg Unversy CenER and deparmen of Accounancy Ths draf: December 10, 2007 JEL-classfcaon: G15, K22, M41 We acknowledge he helpful commens of Phl Berger, Peer Eason, Suar Mcleay, Rchard Macve, DJ Nanda, Peer Pope, Lakshmanan (Shva) Shvakumar, Balj Sdhu, Jeroen Sujs, Rodrgo Verd, Ross Was and workshop parcpans a Tlburg Unversy, he Caholc Unversy of Leuven, he 2007 European Accounng Assocaon Annual Conference (Lsbon), he 2007 Mancheser Busness School CAIR Conference and he 2007 London Busness School Accounng Symposum. The auhors acknowledge fnancal suppor from he European Commsson Research Tranng Nework INTACCT (MRTN-CT ). 1

2 Absrac We nvesgae he comparably of accounng earnngs for EU counres n he perod Alhough pror sudes have documened nernaonal dfferences n fnancal reporng properes, here s hardly any large-scale nernaonal evdence on he underlyng fundamenals drvng hese dfferences. The EU s a parcularly neresng seng because of he effors o creae a sngle marke. We focus on he accruals cash flow assocaon as n Ball and Shvakumar (2005; 2006) and show ha accruals measuremen s subsanally affeced by he busness cycle sage and frm specfc reporng ncenves. Incenves arse from he equy capal marke, deb fnancng and labor markes. These ncenves are nensfed by a counry s nsuonal framework, such as sock marke developmen, mporance of bank fnancng and labor unon membershp. In addon, our resuls sugges ha he mandaory nroducon of IFRS n 2005 dd no nsanly brng abou he expeced mprovemen n earnngs comparably across Europe. Our resuls conrbue o he ongong debae on sandards versus ncenves. 2

3 1. Inroducon In hs sudy, we nvesgae he deermnans of nernaonal earnngs comparably over me. Conssen wh regulaor vews (e.g., Jenkns 1999) and pror academc research (e.g., Land & Lang 2002), we defne reporng comparably as he ably of earnngs o accoun smlarly for alke ransacons and dfferenly for dssmlar ransacons. Gven he ncreased level of cross-border ransacons over he las few decades and he hsorcal dfferences n local Generally Acceped Accounng Prncples (GAAP), mprovng comparably has become a major challenge for sandard seers. I should herefore no come as a surprse ha boh he FASB and IASB pu consderable emphass on defnng and mplemenng he concep of comparably n her laes jon Concepual Framework: [Comparably] s a key characersc of decson-useful fnancal nformaon, as would enable users o denfy smlares n and dfferences beween wo ses of economc phenomena (Crooch, 2006). In Aprl 2007, he SEC launched he dea o allow US domesc frms (as well as foregn regsrans n he US) o use IFRS nsead of US GAAP. Ths recen revoluonary SEC proposal emphaszes our undersandng of he level and deermnans of comparably. In our sudy, we pos ha nernaonal earnngs comparably assumes boh ransacon smlary and smlary n accounng pracces as refleced n he accrual accounng sysem. Pror nernaonal accounng research has employed varous mercs o examne crosscounry dfferences n earnngs characerscs: earnngs mulples, accruals-cash flow assocaons and he degree of conservasm (Pope and Rees 1992; Joos and Lang 1994; Gner & Rees 2001; Land and Lang 2002). Smlar o Ball and Shvakumar (2005; 2006), we focus on he accrual accounng sysem o nvesgae he assocaon beween accruals and posve, resp. negave cash flows across dfferen counres. We conjecure ha earnngs 3

4 comparably s largely affeced by he way he accruals sysem recognzes losses n a mely fashon or smoohes ncome over dsnc reporng perods. In addon, we sudy he sensvy of hs accruals cash flow assocaon o frm-specfc reporng ncenves and busness cycles across a se of counres. By dong so, we compare he way accrual accounng funcons equally (or, dfferenly) across counres, whch allows us o nerpre he comparably of nernaonal earnngs. We choose a sample of European Unon (EU) frms over a 15 year perod pror and up o IFRS nroducon ( ) as our research seng for wo mporan of reasons. Frs, snce he mplemenaon of he Sngle Marke Program n 1992, he European economc envronmen became consderably more smlar due o he abolshmen of economc and proecons barrers. Snce hen, macro-economc polces (such as currency uns, neres raes and fscal polces) are se by he EU Commsson nsead of a he counry level, resulng n more negraed Member counres economes. Moreover, cross-border compeon n Europe has ncreased consderably snce he early 1990s (Adjaoué and Danhne, 2004). 1 As a resul of hs ncreased cross-border compeon and more negraed economes snce he early 1990s, we expec o observe more comparable fnancal reporng and smlar effecs of reporng ncenves on he accrual measuremen sysem. Second, n addon o he economc negraon, Europe also s he frs large-scale conduc es of a radcal mandaory swch of naonal, counry-specfc accounng sandards owards unform Inernaonal Fnancal Reporng Sandards (IFRS). Despe he early 1980s naves o harmonze accounng polces across Europe wh he 4 h and 7h Drecve, 1 More smulaneous busness cycles and cross-border compeon hreas suggess ha ndusry classfcaons a he supranaonal level have become more mporan han pure counry barrers, especally for EU counres (Baele and Inghelbrech, 2006). 4

5 naonal GAAPs sll connued o vary across counres (Joos and Lang, 1994). Sree (2002) shows ha hese dfferences sll perss n naonal EU regulaons for more recen years. Moreover, whle some EU frms volunarly repored accordng o nernaonal sandards (IFRS) before he mandaory adopon year 2005, we show ha hs s only a margnal rend n he EU conex, and s lkely nsuffcen o shape average counry reporng phenomena. Wh he mplemenaon and planned ancpaon of IFRS, however, he accruals measuremen sysem becomes unform across all Member counres, whch n urn subsanally reduces counry-specfc varaons n accruals measuremen. Conssen wh convergng accounng sandards, we expec o fnd more smlar accruals cash flow assocaons across counres over me. However, remanng dfferences n reporng ncenves across counres (despe he economc negraon effors) mgh dmnsh he ncrease n earnngs comparably. 2 As such, he EU seng s well sued o sudy he role of ncenves versus sandards over a longer me perod n boh elemens could change. To emprcally nvesgae he role of reporng ncenves n explanng cross-counry dfferences n he accruals - cash flow (hereafer, CFO) relaon, we exend he pecewse lnear regresson model developed by Ball and Shvakumar (2005; 2006) separang posve and negave cash flow observaons. We nclude reporng ncenves relaed o () capal marke pressure, () deb levels, and () labor relaon ncenves, whle conrollng for busness cycles. 3 As suggesed by Krschenheer and Melumad (2002), capal marke 2 Ths resul would be n lne wh recen anecdoal evdence suggesng ha dfferences n fnancal reporng perss afer he 2005 mandaory IFRS nroducon (Erns & Young 2006). Based on a lmed sample of he 100 larges EU non-fnancal frms, an Erns and Young (2006) research memorandum concludes ha fnancal saemens of a French realer, for example, look and feel more smlar o hose of a French manufacurer han o hose of a Duch or UK realer. 3 Noe ha Ball and Shvakumar (2005) nclude sze, leverage and ndusry conrols n her regresson model o conrol for dfferences n he exogenous volaly n economc ncome (p. 110) alhough her man focus of analyss s on he separaon beween publc and prvae UK frms. Par of our analyses, however, rely on 5

6 pressure could nduce an accrual reporng behavor where smoohng and bg bah accounng co-exs. However, Dechow and Sknner (2000) nerpre capal marke pressure as ncreased managers sensvy o sock prce volaly, whch could explan oppose expecaons of mely gan and unmely loss recognon. The ncdence of hgh deb levels wll lkely resul n an ncreased demand for more mely loss recognon, especally n recen years when EU lendng acvy has nernaonalzed sgnfcanly and credors have become more acve monors (Degeorge and Maug, 2006; Ball e al., 2007). Fnally, labor pressure s an mporan feaure of EU socal democraces, far more han has been n a ypcal US seng (Roe, 2003; Ball e al., 2000). We conjecure ha labor pressure encourages smoohng of posve cash flow as well as mely loss recognon as hs reduces prof argumens n collecve wage barganng or lay-off decsons (Lbery and Zmmerman, 1986). In addon, we paron our observaon perod accordng o busness cycle regmes as suggesed by Ryan (2007) o sudy he dfferenal mpac of reporng behavor across economc recessons and expansons. We fnd consderable dfferences n accrual measuremen across he EU counres over he perod Posve cash flows are offse for 72% by ncome decreasng accruals (.e., smoohng) n he early 1990s whle for 62% afer Smoohng of posve cash flows seems o be a general feaure across Europe, bu wh greaer dfferences across counres n laer perods. In he early 1990s, mos counres (excep for Denmark, Ireland and UK) do no seem o have an asymmery beween he recognon of losses and gans, as measured by negave respecvely posve cash flows. More counres, however, exhb mely loss recognon n more recen years; a fndng conssen wh a greaer degree of earnngs ncenve effecs proxed for by sze, leverage and labor pressure as szeable frm-level ncenves ha do no smply capure exogenous volaly n economc ncome. 6

7 conservasm and herefore hgher earnngs qualy. However, perssen smoohng of losses characerzes reporng pracce n counres such as Greece and Porugal. Excludng hese wo counres from he analyss reveals a more comparable and more mely loss recognon across he EU. These fndngs sugges lower earnngs comparably across he EU n more recen years when posve cash flow frms are nvolved and greaer comparably when negave cash flow frms are nvolved. Nex, we nvesgae how reporng ncenves affec accrual measuremen and wheher he ncenve effecs dffer across he EU over he perod We show ha capal marke pressure s assocaed wh less smoohng n posve CFO years and less mely loss recognon n negave CFO years, and especally so for France, Germany and he UK. Ths s evdence conssen wh he capal marke pressure hypohess of Dechow and Sknner (2000). Furher, we show ha frms wh hgh deb levels smooh posve CFO more across all perods and recognze losses more mely, bu manly so n he recesson perod. Ths fndng s n lne wh an ncreased demand for more conservave reporng, especally durng economc recesson perods. In addon, we fnd ha labor pressure s on he one hand posvely assocaed wh smoohng of posve profs across all EU counres, and on he oher hand assocaed more mely loss recognon, especally n he recesson perod. Ths fndng confrms our expecaons ha for EU counres, labor pressure s an mperave deermnan of reporng behavor and ha labor-nensve frms reduce he mpac of collecve wage barganng by keepng repored profs as low as possble (Lbery and Zmmerman, 1986). Fnally, hese combned resuls also sugges ha alhough ncenve effecs vary over me and across counres, hey do no dsappear as from 2005 (.e., he mandaory IFRS adopon year). 7

8 We check he robusness of our resuls wh respec o conrols for secor specfcaon, growh characerscs as well as for ncludng marke reurns n our pece-wse regresson model. Alhough secor separaon explans some varaon n nernaonal accruals measuremen, s no responsble for he observed accruals CFO evoluon across counres over me. In lne wh Dechow e al. (1998), we denfy frm-specfc growh characerscs as a more mporan facor of dfferences n nernaonal earnngs comparably. Fnally, we fnd ha he book-based accruals CFO assocaon s no sgnfcanly alered by ncludng a marke-based proxy for economc gans and losses. Fnally, we sudy wheher hree nsuonal counry feaures (mporance of sock marke, domesc bank deb, and labor unon membershp) affecs he observed effec of frm reporng ncenves on (un)mely loss/gans recognon. Our resuls show ha he nsuonal feaures of a counry ndeed affec he frm reporng ncenves, and he effec depends on he busness cycle. In 2005, here s no srong nsuonal effec anymore on he reporng ncenves, possbly suggesng he resul of converged nsuonal feaures n he EU. In sum, our resuls sugges ha n an nernaonal conex, economy-wde fundamenals ogeher wh ndusry-specfc facors as well as frm-specfc ncenves and nrnsc frm growh jonly mpac he properes of accounng earnngs. We fnd hese paerns boh for he perods before as well as n he frs year of mandaory adopon of IFRS. Snce frmspecfc ncenves reman mporan drvers of he reporng behavor, we argue ha ncenve effecs mgh connue playng a domnan role n accruals measuremen, even afer nroducng one common se of accounng sandards. Ths fndng s n lne wh earler fndngs by Ball e al. (2003) and suggess ha a mere swch o a common se of accounng 8

9 sandards (IFRS) s no necessarly suffcen o brng abou he ancpaed mprovemen n earnngs qualy or can enhance earnngs comparably whn Europe. Our sudy provdes a number of nsghs for he nernaonal accounng leraure. Frs, we show ha nernaonal convergence a he macro-level as well as polcal harmonzaon effors lead o ncreased nernaonal earnngs comparably paerns, despe he applcaon of dfferen accounng sandards a he counry-level. Second, we show ha frm-specfc ncenves relaed o capal and deb markes as well as labor pressure a he counry level are mporan deermnans of he observed varaon n nernaonal earnngs. In addon, our fndngs sugges ha ncenve effecs connue o deermne varaons n accruals measuremen, even afer he mandaory mplemenaon of IFRS. Ths suggess ha a mandaed swch o one common se of accounng sandards s no necessarly suffcen per se o brng abou a cohesve mprovemen n nernaonal earnngs comparably. The remander of he paper s organzed as follows. In he second secon, we descrbe he European envronmen, summarze he mos mporan economc and polcal changes and buld on hese feaures and he relaed leraure o buld our hypoheses. Secon 3 descrbes he daa and sample composon and provdes descrpve sascs on our sample. In secon 4, we presen our emprcal fndngs and we provde addonal ess n Secon 5. In he fnal secon, we summarze and conclude he sudy. 2. The European conex Durng he second half of he 20 h cenury, he macro economc envronmen n Europe experenced a dramac ransformaon, as many aemps were se up o economcally and 9

10 polcally une several counres no a sngle European Unon (EU). One of he mlesones n hs unfcaon process was he esablshmen of a Sngle Marke Program n 1992, oulned n he Maasrch Treay, whch allowed free goods and servces rades hroughou he EU member saes. Ths mporan deregulaon of produc, servces and labor markes sgnfcanly promoed rade beween frms from dfferen EU member saes (Adjaoué and Danhnne, 2004). Moreover, he Maasrch Treay also se he goals for esablshng a common moneary polcy monored by a Cenral European Bank, nroducng a sngle currency (he Euro) for all Member Saes, and algnng fscal polces based on a common dscplne. Ths process of economc and polcal negraon s unque from a global perspecve and has been he subjec of a rch leraure n many economc dscplnes lke fnance, ndusral economcs and accounng. A he level of fnancal markes, Bekaer e al. (2007) repor a sgnfcan upward rend n sock reurn correlaons across Europe snce he early 1990s, conssen wh progressvely more negraed EU fnancal markes. Oher sudes (e.g., Baele and Inghelbrech, 2006) demonsrae ha srucural facors such as he removal of rade barrers or he nroducon of a sngle currency sgnfcanly reduce counry-specfc volaly n sock reurns. Gerosk and Gugler (2004) documen ha despe he EU economc negraon n he 1990s, naonal EU marke srucures sll exhb dssmlar marke-sze srucures. Conssen wh hese fndngs, Goddard e al. (2005) provde n a four-counry EU sudy (Belgum, France, Ialy and he UK) evdence of perssen excess profably realzaons over he perod , despe he exsence of he Sngle Marke Program. Apar from varous aemps for economc negraon, he European Commsson [EC] also made varous aemps o harmonze he fnancal reporng scene n he EU over he las 40 10

11 years. A frs aemp for reporng harmonzaon ook place hrough he nroducon of European accounng drecves. A drecve s a legal nsrumen, whch each Member Sae has o nver no naonal law whn a ceran perod of me. The Fourh and Sevenh Drecve were nroduced n he 1980s as gudelnes o harmonze naonal law wh respec o accounng rules across all Member Saes. However, he expeced harmonzaon effec was nsuffcen and he EC swched course n he 1990s owards he prescrpon of rules n lne wh he Inernaonal Accounng Sandards (IAS/IFRS). When he EC realzed n he 1990s ha he harmonzaon was nsuffcen, ook unl 2002 o approve a gudelne requrng all EU lsed frms o repor n accordance wh IFRS as from he fscal year 2005 (EC Regulaon No. 1606/2002). Well before he approval of hs gudelne n he European Parlemen, some ndvdual EU frms already swched volunarly owards IFRS (e.g., Vansraelen and Van Tendeloo, 2005). Table 1 provdes an overvew of he volunary IFRS adopers n he EU snce 1995, when he Inernaonal Accounng Sandards Board (IASB) sared s core sandards program. In 2005 all EU member saes adoped IFRS, wh he excepon of Londen AIM frms ha can sll use local GAAP n We nfer from hese fgures ha a very low proporon of frms are classfed as volunary IFRS adopers before The only excepons n he EU-15 area are Ausra and Germany, wo counres ha explcly oblged lsed frms o repor accordng o an nernaonal sandard, eher IFRS or US GAAP. For all oher counres, reporng accordng o IFRS (alone) was no allowed, or when allowed only pracced by very few companes. 4 4 Noe ha fnancal reporng n accordance wh IFRS alone was no allowed before he fscal year 2005 n 6 ou of of our sample counres, namely France, Ialy, he Neherlands, Span, Sweden and he Uned Kngdom ( IAS: use of opons). 11

12 INSERT TABLE 1 ABOUT HERE In nernaonal accounng sudes, he EU conex and s economc negraon process has also lend self o varous sudes. In an early sudy, Joos and Lang (1994) show ha crosscounry accounng measuremen pracces affec he comparably of nernaonal fnancal raos n he perod Land and Lang (2002), however, documen convergence rends n earnngs-o-prce mulples n a more recen perod ( ) across 7 major jursdcons worldwde, amongs whch hree are EU counres: France, Germany and he UK. Oher sudes (e.g., Gner and Rees, 2001; Raonc e al., 2005) show ha condonal accounng conservasm vares subsanally across EU counres and s lkely affeced by reporng ncenves such as legal envronmen and capal markes mporance. In a more recen sudy, Burgsahler e al. (2006) documen ha legal envronmen also domnaes earnngs managemen relaons a he counry level. Despe conflcng vews on EU convergence and economc negraon effecveness across and even whn dfferen research domans, general macro-economc and nsuonal sascs do provde evdence n lne wh negrang EU economes durng he pas 15 o 20 years. Fgure 1 summarzes a number of macro-economc and nsuonal dsperson rae evoluons across EU counres from he early 1990s unl We calculae dsperson n nflaon raes, GDP per capa and unemploymen raes as macro-ndcaors and dsperson n sock marke capalzaon per capa, average deb o equy levels and he proporon of foregn sales o oal sales as nsuonal varables 5. For all bu one dsperson evoluon 5 Dsperson s defned as he coeffcen of varaon of he varable of neres across all EU Member Saes. 12

13 (GDP/capa), we observe a clear paern of less dsperson when me moves on. Decreasng dsperson rends are mos pronounced for nflaon raes (from 72 o 39 per cen), marke capalzaon mporance (from 71 o 40 per cen) and relave nernaonal sales (from 68 o 32 per cen) INSERT FIGURE 1 ABOUT HERE Ths EU seng where marke enry barrers and rade regulaons have been subsanally reduced, provde a unque seng o sudy paerns n nernaonal earnngs comparably. Moreover, he conex s unque n ha accounng prncples dffered subsanally a he me of he nroducon of he Sngle Marke Program and ha sgnfcan aemps have been made o sandardze hs reporng n more recen perods. 3. Sample, Daa and Descrpves We use fnancal saemen daa from he Thomson Worldscope daabase for he perod The nal sample consss of all non-fnancal companes ncorporaed n one of he 15 EU Member saes (.e., he EU Member counres as before May 2004), and havng a prmary lsng n her respecve home counry. We exclude Luxembourg because of nsuffcen frm-year observaons. We also exclude frm-year observaons wh mssng Indusry Classfcaon Codes (ICB), wh a fscal year exceedng 12 monhs, or wh mssng daa on oal asses (WC02999), earnngs (WC01706), or on accruals componens. 13

14 Conssen wh recen sudes n nernaonal fnance (Baele and Inghelbrech, 2006; Bekaer e al., 2007) we use he Indusry Classfcaon Benchmark (ICB) secor codes, also known as he FTSE Acuares Classfcaon Sysem o denfy ndusry membershp. Ths ndusry classfcaon benchmark ouperforms a ypcal SIC-classfcaon when sudyng non-us socks (Bekaer e al., 2007) by classfyng ndvdual frms n a vas range of 39 homogeneous ndusres. We exclude he 7 fnancal secors from our sample resulng n a fnal secor classfcaon of 32 non-fnancal secors (See Appendx 1 for more deals). We conduc he analyses over 4 dfferen perods, each reflecng dsnc macro-economc shocks: (perod 1) refers o he early EU negraon perod and was a recesson perod; (perod 2) by conras was characerzed by boomng marke condons (expanson), whereas (perod 3) s agan denfed as a recesson perod (CEPR Sascs; See Fgure 2). 6 Fnally, he 2005 observaons agan are from an expanson perod and have he unque feaure ha all repored earnngs are n accordance wh one unque nernaonal sandard, namely IFRS. We conrol for busness cycles o address concerns ha accounng fgures and especally losses are poenally relaed o he busness cycle (see e.g. Klen and Marquard, 2006; Ryan, 2007). Klen and Marquard show ha he propensy of accounng losses s posvely relaed o economc recessons. Khurana e al. (2006) documen ha asymmerc mely loss recognon declnes durng perods when he economy s n a relavely srong phase, conssen wh he hypohess ha he more adverse nvesor response o earnngs 6 Noe ha CEPR (Cener for Economc Polcy Research) s he European equvalen of he US NBER (Naonal Bureau for Economc Research). CEPR provdes monhly ndcaors of he EU area busness cycle on a conssen bass snce he lae 1980s. More nformaon on he CEPR and he recesson expanson ndcaor can be found n Fgure 3.

15 dsapponmens durng prosperous mes creaes an ncenve for managers o lower he asymmerc melness. We nclude he 2005 observaons as a separae perod o examne he effec of he mandaory accounng sandards swch on nernaonal earnngs comparably. As from January , EU lsed frms are requred o repor consoldaed fnancal saemen n accordance wh IFRS (Inernaonal Fnancal Reporng Sandards). 7 Whereas pre-2005 earnngs are poenally sll dfferenly affeced by varaons n local GAAP prescrpons defned a he counry level, 2005 IFRS fgures should no be based as such f all counres apply he sandards n a smlar and conssen way. Earnngs are decomposed no accruals and cash from operaon. Followng Dechow e al. (1995) and Leuz e al. (2003), we compue accruals as: (ΔCA - ΔCASH) - (ΔCL - ΔSTDEBT) DEPR, where ΔCA s he change n curren asses (WC02201) from year o -1, ΔCASH s he change n cash (WC02001), ΔCL s he change n curren lables (WC03101), ΔSTDEBT s he change n shor erm deb (WC03051), and DEPR s he annual deprecaon, depleon and amorzaon expense n year (WC01151). Deducng accruals from earnngs provdes us wh an operang cash flow measure (hereafer CFO). To allow for drec frm sze comparsons, we deflae earnngs, cash from operaons and accruals by oal asses. Our fnal sample consss of 25,110 observaons across he 4 perods, coverng counres and 32 non-fnancal ICB secors. 7 As from January 1, 2005 he mandaory adopon of IFRS became effecve for he majory of EU lsed frms. UK frms lsed on he Alernave Invesmen Marke (AIM) are, however, able o pospone he adopon of IFRS wh wo more years unl he fscal year Furher excepons exs and ypcally nclude frms currenly already reporng reconclaed fgures under US GAAP (.e. EU-based frms wh cross-lsngs n he US). 15

16 INSERT TABLE 2 ABOUT HERE In Table 2, we presen he mean, medan and sandard devaon of 7 varables of neres:.e. () he 3 key varables (earnngs, cash from operaons and accruals), () 3 reporng ncenves (marke capalzaon, fnancal leverage, and labor nensy), and () he hghech ndcaor varable. The reporng ncenves are our man varables used n he mulvarae secon of he analyses and hgh ech s our dchoomous secor conrol. The frm-year observaons correspond wh hese used n he regresson analyss repored n Table 4. Oulers have been deleed by applyng a mulvarae deleon rule where he op and boom 1 per cen of observaons are no ncluded n he regresson (see Table 4). 8 Earnngs seem o follow he busness cycle paern, wh a negave average value of 3.1% n he recesson of Ths earnngs paern s dfferen from he seady declne documened for he US by Gvoly and Hayn (2000, p.296), n Darrough and Ye (2007, p. 36), and Klen and Marquard (2006, p. 185). Gvoly and Hayn arbue he declne n earnngs o an ncreased conservasm whereas Darrough and Ye explan hs paern by he ncreasng profably gap beween large and small lss. Klen and Marquard provde evdence ha he declnng average profably s also relaed o underlyng economc performance (as measured by CFO). An eye-cachng rend n Table 2 s he seady ncrease n he sandard devaon of earnngs n he frs hree perods followed by a declne n Ths ncrease n earnngs varably 8 Resuls are unaffeced f we rese hs ouler deecon rule o 0.5 and 2 per cen. 16

17 unl he early 2000s s n lne wh an ncreasng CFO varably bu sands n conras wh accruals paerns ha remaned far more sable over me. Comparng he mean and medan values of earnngs and CFO show ha hese dsrbuons are heavly skewed, especally n he recesson perod where exreme negave earnngs and CFO bas he mean values downward. As descrbed above, we have 32 non-fnancal secors ncluded n he sample. Conssen wh Hayn (1995) and Hand (2005) we fnd n unrepored analyses (avalable upon reques) ha hgh-ech frms follow sysemacally dfferen earnngs paerns han low-ech ndusres across hese 32 secors. A frm s consdered o be hgh-ech (a dummy varable equal o 1 f he frm belongs o a hgh-ech ndusry and zero oherwse) f belongs o one of he followng 6 ICB secors: Aerospace & Defense (#271), Pharma & Boech (#457), Fxed Lne Telecom (#653), Moble Telecom (#657), Sofware & Compuer Servces (#953), Technology Hardware & Equpmen (#957). Table 2 shows an ncrease n hgh-ech frms from 7.5% n o 22.5% n 2005, manly drven by an ncrease n Sofware and Compuer Servces frms. Paerns n earnngs and properes of he earnngs componens wll be relaed o frm-specfc reporng ncenve varables n he mulvarae analyses. We herefore provde addonal descrpve sascs on marke capalzaon (SIZE), fnancal leverage (measured as long erm deb as a proporon of oal asses) and labor nensy (measured as number of employees per mllon EURO marke capalzaon) over me. 9 The average frm sze 9 We choose o scale he number of employees by oal marke value as our esmae for labor nensy. Alernavely, one could choose o scale by oal asses or sales. Alhough neher marke value nor asses or sales are an opmal scale measure, we choose o use for marke value snce hs value conans boh realzed and unrealzed growh opporunes whch are no necessarly refleced n sales or oal asses. We gve he example 17

18 follows he general busness cycle rends wh larges values n perod 2 ( ) and Fnancal leverage s farly sable over me around 20 per cen of oal asses. Fnally, frms become ncreasngly less labor nensve over me; a fndng conssen wh he ncreasng mporance of hgh-ech, knowledge-based hough less labor nensve frms INSERT TABLE 3 ABOUT HERE Afer he dscusson of he pooled sample descrpve sascs, we repor he counry specfc averages of he varables. Average earnngs levels no only vary subsanally across me, bu also across counres. The relave levels of earnngs for France, Germany and he UK are conssen wh earler fndngs by Joos and Lang (1994) for he 1980s. German frms conssenly have low levels of earnngs. UK earnngs are larger han hese of France and Germany, bu only unl 2000 when average earnngs drop o smlar levels as Germany n he perod Furher, Swedsh frms record he larges losses n he recesson perod, and relavely low earnngs levels n he frs and las perod. The subsanal accounng losses n Sweden are generaed hrough negave cash flows and nsananeous negave accruals. In conras, Greek frms exhb a relavely low level of negave accruals and posve cash flows, resulng n relavely sable earnngs over me. In general, hese resuls for Greece do no follow he same rend as n he oher European counres (especally n ). of a boech frm havng vrually no or very low sales (respecvely, asses) o llusrae ha wh sales or oal asses as denomnaor, some frms could be ranked as beng very labor nensve smply because of her small denomnaor value. 18

19 Furher, more han 8% of he early 1990s lsed frms n Denmark (9.3%), Fnland (12.9%), France (9.4%), Ialy (8.5%) and he Neherlands (8.9%) are hgh-ech frms. Oher counres lke Ausra and Porugal dd no have any hgh-ech frms n he early observaon perod. By 2005, however, all EU counres have more han 10% hgh ech frms, wh Fnland (28.7%), Germany (29.3%) and Sweden (33.7%) havng around 30% of her frms n he hgh-ech secors. The average marke capalzaon vares subsanally across he EU n all 4 perods. The average lsed frm sgnfcanly grows over me, especally so n Porugal (by 16.5 mes), Ireland (by 8.5 mes) and Ausra (by 5.8 mes) from perod 1 o perod Fnancal leverage s ncreasng n Greece (from 0.22 o 0.32) and Porugal (from 0.26 o 0.41), and sgnfcanly decreasng n Sweden (from 0.31 o 0.17) and Fnland (from 0.45 o 0.26). The rend n leverage s nversely relaed wh he rend n hgh-ech frms n hese wo Scandnavan counres, conssen wh hgh-ech frms relyng more on equy fnancng. Labor nensy s sgnfcanly decreasng over me, excep n Greece (from 15.7 o 12.7 employees as per mllon Euro marke cap) and o a lesser exen n Germany (from 24.0 o 16.4 employees as per mllon Euro marke cap). Sweden has he mos specacular decrease n labor nensy from 41.6 o 7.5. Labor nensy vares beween 15.7 (Greece) and 44.2 (France) n and beween 6.2 (Span) and 16.9 (France) n One should be cauous when nerpreng he marke capalzaon rends, snce he dsrbuons are heavly skewed and he repored averages are affeced by a few exremely large frms. Noe already ha he presence of hese exremes do no affec or conamnae furher es resuls as we wll use he sandardzed (0,1) rank of marke capalzaon, fnancal leverage and labor nensy n our mulvarae ess. 19

20 4. Mulvarae Analyses 4.1 Accruals Cash Flow Relaons, Asymmerc Tmelness of Loss Recognon and Busness Cycle Sages In hs secon, we nvesgae he evoluon of he asymmerc relaon beween accruals and posve (resp. negave) cash flows over a 15 year perod across EU counres. By dong so, we hghlgh he (poenally) changng role of accrual accounng n performance (earnngs) measuremen over me across he EU counres. More specfcally, Dechow e al. (1998) argue ha accruals mgae nose n operang cash flow, and herefore exhb a negave correlaon wh operang cash flow. Ball and Shvakumar (2005; 2006) dscuss a second role of accrual accounng, whch s he mely recognon of gans and losses. They show ha a more mely recognon of economc gans or losses nduces a posve correlaon beween accruals and cash flows. The more conservave an accounng sysem s, he greaer he melness of loss recognon versus gans recognon and he greaer he degree of asymmery. The model we apply here s an exenson of he pecewse lnear accruals-cash flow model developed by Ball and Shvakumar (2005): ACC = α = 1 1, α DPCFO + = 1 1, β DPCFO CFO + α = 1 2, DNCFO = 1 2, β DNCFO CFO + ε (2) Where ACC s equal o oal accruals, CFO are operang cash flows, DPCFO (DNCFO) s a dummy varable equal o one f cash flow s posve (resp. negave) and DPCFOxCFO (DNCFOxCFO) s equal o he neracon beween he dummy varables and he cash flow values. 20

21 The coeffcens n regresson model (2) can vary across he EU counres. The coeffcen β 1 provdes a counry-specfc paral correlaon coeffcen of he ACC - CFO assocaon provded ha cash flows are posve. The coeffcen on β 2 hen represens he counryspecfc accruals cash flow paral correlaon coeffcen for negave cash flow years. Asymmery beween loss and gan recognon n he model resuls n a β 1 coeffcen ha s more negave han he β 2 coeffcen. Smlar o he analyses n Table 2 and 3, we separaely esmae regresson model (2) for he four sub-perods as defned above ( [1]; [2]; [3]; 2005 [4]) o conrol for busness cycle effecs INSERT TABLE 4 ABOUT HERE Resuls n Table 4 show ha across all observaon perods, he accruals cash coeffcen s on average more negave for posve (β 1 ) han for negave cash flow (β 2 ) frm-years. However, whereas he dfference s relavely small n he frs perod ( : 71.8 for posve CFO versus 56.8 for negave CFO), hs dfference subsanally ncreases for recesson years ( 61.9 for posve CFO versus 5.8 for negave CFO), conssen wh greaer asymmery n gan/loss recognon over me. The greaer asymmery n more recen years s caused by a sgnfcan ncrease n mely loss recognon n recen years versus a moderae ncrease n mely gans recognon. Porugal and Greece are he excepons and show no dfference n accruals-cash flow relaon beween posve and negave cash flows. These fndngs of ncreasng loss recognon asymmery over me are conssen wh an ncrease n earnngs qualy beween 1990 and Land and Lang (2002) also fnd accruals-cash flow correlaons o become less negave n he perod

22 and 1999 and nerpre hs resul as evdence of less earnngs smoohng. However, our pece-wse lnear model emphaszes he mporance of separang ou negave versus posve CFO n he accruals-cfo relaon. Below we furher explore he role of busness cycles and reporng ncenves as drvers of he accruals CFO assocaon. We observe subsanal counry dfferences n he β 1 coeffcen (ACC-posve CFO relaon) n all perods, and counry dfferences ncrease over me as he sandard devaon of he counry coeffcens monooncally ncrease from 7.7% n perod [1] o 19.6% n 2005 (IFRS year). We furher nvesgae counry dfferences across busness cycles. In perod (a recesson perod), Ireland (-54.2) and he UK (-64.9) exhb he leas negave correlaon for posve CFO, conssen wh earler research fndngs showng lower earnngs smoohng and herefore hgher earnngs qualy compared o connenal European frms (e.g., Joos and Lang, 1993). Counres such as Belgum, Fnland, Sweden and Span sar showng a less negave β 1 coeffcen n he perod (expanson). Furher, we fnd ha Germany keeps a hghly negave accruals - CFO correlaon coeffcen β 1 n all four perods. Ths mgh come as a surprse, gven ha many German frms are early IFRS adopng frms bu confrm earler resuls as n Van Tendeloo and Vansraelen (2005). The dfferences n he β 2 counry coeffcens (accruals-negave CFO relaon) are hgh and farly sable n all perods (sd [1] = 42.8% versus sd [4] = 39.1%). In he frs recesson perod ( ), Ireland (25.6), he UK (2.1) and Denmark (10.3) have a posve accrual-negave CFO relaon, ndcang a mely recognon of economc losses. The oher EU counres exhb smlar smoohng behavor as n he case of posve CFO-accruals correlaon,.e. a negave β 2. Sarng from he second perod ( ), negave cash flows are much less compensaed by ncome ncreasng accruals n France, Germany, Ireland, Sweden and he 22

23 UK. Agan, Greece and Porugal are noceable excepons havng sgnfcanly negave β 2 coeffcens n all perods. All oher EU counres, however, exhb more mely loss recognon n more recen perods compared o he early 90s. Furher, we fnd ha for 2005 (expanson perod and he frs year of mandaory IFRS adopon), β 2 coeffcens on average become more negave agan. 11 We nerpre hs as evdence conssen wh ncreasng melness of loss recognon n economc downurn perods, rrespecve of applyng one sngle se of accounng sandards (IFRS). Ths fndng hghlghs he role of busness cycles n undersandng how accruals work. In economc recessons, he correlaon beween curren and fuure negave cash flows mgh be sronger han n non-recesson or expanson perods. If economc losses are refleced n decreasng fuure cash flows, hen curren negave cash flow s lkely o be posvely correlaed wh fuure cash flows. To he exen he reducon n fuure cash flows are accrued as a componen of curren perod earnngs (for example by an mparmen charge), curren cash flows are posvely correlaed wh accruals (Ball and Shvakumar, 2005). Combned, hese resuls show ha economc losses are recognzed dfferenly across busness perods and ha EU frms recognze losses more mely n he more recen recesson perod as opposed o he early 1990s recesson years. However, on op of he more mely loss recognon n more recen years, he subsanal varaon a he counry level rases quesons abou nnae whn-counry dfferences poenally gearng hese cross-counry dfferences. Mos earler nernaonal accounng sudes ypcally use counry-level sac varables 11 Noe ha coeffcens for negave 2005 CFO observaons canno be esmaed for Ausra, Belgum, Fnland, Ireland and Span because of a lack of suffcen (mnmum 7) daa o run our regresson. We requre a leas 7 observaons per counry o esmae our regresson model. Changng he requred number of observaons o 10 does no aler he general resuls bu does leave ou 3 more counres (Denmark (8), Neherlands (8) and Porugal (10)). 23

24 relaed o rules of law and corporae governance o model counry dfferences n earnngs properes (Ball e al., 2000; Leuz e al., 2003; Burgsahler e al., 2006). We argue ha he nrnsc frm ncenves whn one counry are poenal key drvers of he observed reporng properes a he counry level. We elaborae on hese ncenves n he nex secon. 4.2 Reporng Incenves, Accruals Cash Flows Relaons and Asymmerc Tmelness of Loss Recognon In hs secon, we nvesgae he role of whn-counry reporng ncenves o explan he accruals-cash flow relaon. We do so by ncludng separae ncenve varables n model (2): ACC = α = 1 1, = 1 1, = 1 1, α DPCFO + = 1 1, β DPCFO CFO + α = 1 2, = 1 2, γ DPCFO INCENTV + DNCFO = 1 2, δ DPCFO CFO INCENTV + β DNCFO CFO γ DNCFO INCENTV = 1 2, δ DNCFO CFO INCENTV + ε (3) where DPCFO and DNCFO are dummy varables equal o 1 f CFO s posve (resp. negave), and INCENTV s a varable ndcang he reporng ncenve. Our frs ncenve varable (SIZE) focuses on a frm s capal marke pressure and s measured as he whn-counry rank of a frm s marke capalzaon sandardzed o he nerval (0,1). Krschenheer and Melumad (2002) show ha capal marke pressure mgh nduce a reporng behavor where smoohng and bg bah accounng co-exs n equlbrum. However, over he pas wo decades, capal marke pressures ganed sgnfcan mporance and creaed ncenves o boos earnngs n an aemp o mee and exceed analys forecass (Degeorge e al., 1999; Dechow and Sknner, 2000). Capal marke pressure 24

25 herefore could be assocaed wh early gan and delayed loss recognon makng he accruals CFO assocaon less negave for posve CFO frm-years (.e. less smoohng n case of posve economc ncome) and more negave for negave CFO frm-years (.e. more smoohng n case of economc losses). The second ncenve varable s fnancal leverage - LEVERAGE - measured as he whncounry rank of a frm s fnancal leverage rao (long-erm deb scaled by oal asses) sandardzed o he nerval (0,1). Sweeney (1994) shows ha hgh leverage frms are more nclned o manage earnngs o avod deb covenan volaons. On he oher hand, banks are mporan monors a he frm level n varous European counres and que ofen have a board sea n a frm (Degeorge and Maug, 2006). To he exen ha banks have an mporan voce n he company s board and ha reducng bankrupcy coss s one of her key objecves, banks can acually force frms o repor more conservavely (Beay e al., 2007). We argue ha ncreased nernaonalzaon of EU lendng acvy has amplfed he average credor monorng role sgnfcanly, whch n urn sugges an ncreased demand for more mely loss recognon n more recen years. The hrd ncenve varable s a frm s labor nensy - LABOR INTENS - measured as whn-counry rank of a frm s rao of number of employees scaled by marke capalzaon, agan sandardzed n he nerval (0,1). Bronars and Deere (1991) documen ha collecve barganng s well esablshed n many EU counres. Frms can, however, ake acons o reduce he mpac of collecve barganng on corporae profs for nsance by smoohng corporae earnngs va accruals compensaon (Lbery and Zmmerman, 1986). Because of he well esablshed collecve barganng and because employee pressure n Europe are ypcally much greaer compared o a ypcal US conex, we conjecure ha labor nensy s 25

26 an mporan conrol varable n hs EU seng. Smlar o Blne and Cullnan (2003), we capure labor-relaed reporng ncenves by he degree of labor nensy. The upper rgh columns of Table 4 repor he sgn and sgnfcance levels of each of hese hree neracve ncenve varables, ndcaed by a + f sgnfcanly posve, f sgnfcanly negave and 0 f nsgnfcan; and wheher he ncenve neracon effec absorbs he man effec (ndcaed wh a ). Wh absorpon of he man effec we mean ha a sgnfcan ACC CFO paral correlaon coeffcen (capured by β 1 for posve CFOs and β 2 for negave CFO frm-years) observed n regressons whou a conrol for ncenves becomes nsgnfcan afer ncludng he ncenve effec. When hs s he case, we conclude ha frm-specfc ncenves drve he observed ACC CFO assocaon, more han counryspecfc GAAP reporng choces. We fnd ha capal marke pressure has a moderang effec on he negave accruals cash flow assocaon for posve cash flow years. Ths s conssen wh he conjecure ha larger frms experence hgher capal marke pressures o repor gans more mely. Ths posve effec of sze s presen for all perods n Denmark, France, Germany, Greece and he UK. For negave cash flow frms, sze neracons are nsgnfcan for mos of he counres n nearly every perod. For counres such as France, Germany, Sweden and he UK, however, we fnd ha large frms wh negave cash flow recognze losses less mely compared o negave cash flow small frms. Boh resuls are conssen wh he capal marke pressure argumen as n Dechow and Sknner (2000),.e. larger frms smooh losses and quckly recognze gans. The smoohng of losses by larger frms seems o connue n he year of mandaory IFRS adopon. However, larger frms seem o have a smlar paern n mely gans as smaller frms n half of he EU counres, a fndng conssen wh he mprovemen of mely 26

27 gans recognon of smaller frms n counres such as Sweden, Span, Porugal and he Neherlands (denoed by 0 n perod 4 of SIZE). Leverage has a sgnfcanly negave neracon effec for posve cash flow frms, suggesng ha hghly levered posve cash flow frms n almos any EU counry smooh earnngs more han low leverage frms. Ths fndng s conssen wh hghly levered frms lowerng earnngs durng posve cash flow years o buld cooke jar reserves ha can be used n fuure, less posve years. By dong so, frms are more able o avod poenal volaons of her deb covenans (Sweeney, 1994). For negave cash flow frms, he esmaed ACC-CFO coeffcens are mosly nsgnfcan. However, for counres such as Germany, Ialy and he UK he leverage neracon effec s sgnfcanly posve n he mos recen perod and absorbs he man effec. Hgh leverage only has a more mely recognon effec n he UK n recesson perods,.e. we observe a + n perods 1 and 3. Conssen wh he general rend of greaer nernaonal lendng acvy n recen perods, monorng power of nernaonal banks n hghly levered frms poenally has become more mporan resulng n more conservave earnngs reporng. Ths s especally rue n he recesson perod [P3]. Fnally, we fnd ha for nearly all counres and across all me perods, labor-nensve frms have more compensang accruals o reduce posve CFO owards zero. Ths s n lne wh frm acons o reduce labor unons nfluence o negoae on shareholders wealh (Lbery and Zmmerman, 1986). For labor nensve frms wh negave CFO, he accrual cash flow assocaon s less negave (.e. earnngs become equally or even more negave han cash flows) n France, Germany and he UK n all perods. Moreover, hs effec s mos srong n perod [3] and for Germany, labor pressure even drves he mely loss reporng n he frs 27

28 IFRS observaons. Ths fndng s conssen wh he noon ha negave CFO frms are more lkely o consder resrucurng plans. These frms mgh no wan o compensae her negave CFO wh accruals, bu raher wan o show hgh losses o faclae layoff negoaons wh he labor unons. The relaon s especally srong n recesson perods. Our fndngs on he labor nensy effec on he accruals- cash flow relaon sress he mporance of conrollng for labor relaons n sudyng earnngs properes n Europe. Wu and Zhang (2007) show ha he volunary adopon of IAS/US GAAP s relaed o labor relaons n connenal Europe. However, we argue and fnd ha he effec of labor relaons on earnngs properes s more generally presen n Europe. From sudyng reporng ncenve effecs on he accruals-cash flow relaon n he EU over me, we conclude ha reporng ncenve effecs sll dffer across he EU counres, even n he year of mandaory adopon of IFRS, a resul conssen wh Ball e al. (2003). In addon, busness cycles affec he nensy of he reporng ncenve effec on he accrualscash flow relaon hghlghng he ofen-overlooked mporance of busness cycle conrols n accounng properes sudes. 5. Addonal Analyses In hs secon, we es he robusness of our fndngs o alernave model specfcaons. We exend he accruals CFO relaon by addng ndusry membershp, growh conrol and a marke-based ndcaor of economc performance Indusry membershp, Accruals and Cash Flows 28

29 As suggesed by Hayn (1995) and Joos and Plesko (2005), a frm s membershp o hgh ech versus low-ech secors possbly affecs he accruals cash flow assocaon. We herefore nclude HITECH, a dummy varable equal o one f he frm s n a hgh ech secor (as defned n Secon 3 and Appendx 1) and zero oherwse. In Table 5, we repor he esmaed coeffcens of neres from he followng regresson: ACC = α = 1 1, = 1 1, = 1 1, α DPCFO + = 1 1, β DPCFO CFO + = 1 2, γ DPCFO HITECH + α = 1 2, DNCFO = 1 2, δ DPCFO CFO HITECH + β DNCFO CFO γ DNCFO HITECH = 1 2, δ DNCFO CFO HITECH (4) + ε In Panel A, he accruals CFO relaon for posve CFO frms s capured by β 1 (resp. β 1 + δ 1) for low (resp. hgh) ech frms, whle resuls of he accruals CFO relaon for negave CFO frms are presened n Panel B. In general, we observe more mely gan and loss recognon over me: he coeffcen on posve CFO frms drops from -72 n [1] o n [4] for low ech frms and from o for hgh ech frms. Moreover, he asymmery n gan/loss recognon ncreases over me: he β 1 - β 2 dfference s abou 12% n he frs perod and rses o 30.7% n perod [3]. Overall, hese resuls are very conssen wh prevous resuls from Table 4. However, due o he very low number of observaons n ceran subgroups or perods (e.g., hgh ech, negave CFO frms and 2005 daa n parcular) he nerpreaon of resuls s ofen hampered INSERT TABLE 5 ABOUT HERE

30 5.2 Growh, Accruals and Cash Flows We now sudy he robusness of our fndngs o growh characerscs, gven he changng naure of our EU sample whch consss of boh maure and more recenly lsed, hgh growh frms. 12 In parcular, Dechow, Kohar and Was (1998) show ha for hgh growh frms, a sales ncrease (a characersc of growh) frs generaes a ne cash ouflow (because of exendng cred sales and already ncurred producon/servce coss) whereas he ne cash nflow s only realzed wh delay n subsequen perods (when he cred sales are colleced). Ths assocaon hen mechancally nensfes he negave correlaon n year-on-year cash flow changes. To conrol for poenal bas n our man resuls (Table 4) due o growh effecs, we reesmae equaon (2) for each sub perod and addonally nclude a growh conrol, measured as he sandardzed whn-counry frm-rank of asse growh, and nerac hs rank wh counry dummy effecs of posve and negave CFO. 13 For he sake of parsmony, we repor summary sascs of he accruals cash flow coeffcens for each perod across he counres whou growh conrols (lef columns of Table 6) and compare hose wh he coeffcens of (a) negave growh frms (mddle columns) and (b) posve growh frms (rgh columns). 12 In an unrepored analyss, we analyzed he accruals cash flow relaon for a consan sample of frms for whch we have yearly daa n he perod (perods 1 o 3). By focusng on a subse of more maure (lower growh) frms, we dmnsh poenal growh effecs on he ACC - CFO relaon. The resuls are very smlar o full sample resuls bu he number of avalable frm-year observaons drops o 740 resulng n an ofen very low number of counry observaons for negave CFO frms. Exendng he consan sample up o 2005 would furher decrease he sample sze o In unrepored analyses (avalable upon reques), we ncluded he whn-counry sandardzed frm-rank of sales growh as an alernave growh conrol. Resuls are qualavely smlar for boh growh conrols. 30

31 INSERT TABLE 6 ABOUT HERE From panel A of Table 6 we learn ha (posve CFO, negave growh) - frms record on average an accruals cash flow correlaon coeffcen ha s 15 o 20% more negave han for he average frm. Unrepored counry coeffcens show ha hs endency of a more negave correlaon s observed n all EU counres. Negave growh frms ofen experence decreasng accruals caused by lower nvenory levels or accouns recevable. For hese negave growh frms, he cash flow shock as descrbed n Dechow e al. (1998) causes ncreasng CFO levels bu smulaneously reduces workng capal accruals (because of he reducon n nvenory or accouns recevable), resulng n a more negave accruals cash flow assocaon. Furher, he growh effec on he ACC - CFO relaon s sronger n macro-economc downurns ([P1] and [P3]) compared o oher perods. Ths s no surprsng gven ha more frms face srucural cos reducon decsons n workng capal and long erm asses durng recesson perods. The upper rgh par of Table 6 furher shows ha (posve CFO, posve growh) frms exhb a sgnfcanly less negave average CFO coeffcen compared o average growh frms (beween and for posve growh frms compared o 61.9 and 77.8 for negave growh frms). For posve growh frms, posve or ncreasng cred sales concde wh posve or ncreasng cash sales, makng he ACC - CFO assocaon less negave. Noe ha he decson o cu or reduce R&D nvesmens (or sell of propery, plan and equpmen) because of dsapponng growh fgures, would resul n a smlar hghly negave accruals cash flow relaon because of a decreasng long-erm accrual expense (lower propery deprecaon) along wh a smulaneous reducon n cash ouflow (nducng a posve CFO change). 31

32 Panel B of Table 6 summarzes growh effecs on he ACC CFO correlaon for negave CFO frms. For (negave CFO, negave growh) - frms, we show ha he average ACC CFO relaon becomes slghly posve n all four sample perods. These frms are he leas fnancally healhy frms, ypcally facng low or negave profably and deerorang sales erms and suppler cred. These frms face shor and long erm cos reducon decsons, such as nvenory sell ou, reducon of cred sales erms, R&D cus and PPE reducon. These acons cause one-me cash nflows, resulng n a less negave (or even posve) ACC CFO relaon. Fnally, negave CFO and posve growh frms (presened n he rgh columns of Panel B) have a more negave ACC CFO correlaon, whch s agan conssen wh he Dechow e al. (1998) argumen ha posve growh shocks can resul n negave CFO. In general, growh affecs he ACC CFO relaon n predced ways, conssen wh Dechow e al. (1998). Moreover, hese resuls seem o furher confrm he ncenve sory from he prevous secon; growh characerscs a frm level parly shape he ACC CFO assocaon, over and above counry or GAAP dfferences. Fuure research on he ACC CFO relaon whn one counry should herefore furher elaborae on hese fndngs, whle cross-counry sudes wll need o brng he growh facor no accoun. 5.3 Marke Reurn Proxy as an Addonal Proxy for Good versus Bad News In addon o usng he CFO-based accruals proxes as n equaon (2), we re-esmae he condonal ACC CFO model by ncludng sock reurn as a marke proxy for economc gans and losses. To he exen ha sock reurns are a beer proxy for (negave) economc losses han CFO-based measures, sock reurns could be an mporan omed varable n he 32

33 regresson equaon (2). Basu (1997) shows ha (especally negave) sock reurns conan sgnfcan nformaon abou accruals. Ball and Shvakumar (2006), however, show for a sample of US frms ha he marke-based proxy s nferor o book-based proxes for he purpose of denfyng bookable accrued gans and losses. I remans an emprcal queson wheher her fndng holds for European frms INSERT TABLE 7 ABOUT HERE We esmae regresson (2) for each sub perod [1] o [4] separaely and add he year-o-dae sock reurn (RET), a dummy for negave sock reurns (DRET) and he neracon erm DRET*RET. Conssen wh Basu (1997), condonal conservasm s manfesed n he coeffcen on marke-based economc loss (DRET*RET). For he sake of parsmony, we summarze he effec of ncludng he marke-based proxes n our book-based (.e., accruals cash flow) regressons n Table 7. Resuls are very conssen wh he book-based regresson esmaes n Table 4. IN panel A we fnd ha he marke reurn proxy for economc gans does no conan sgnfcan accruals nformaon n he case of posve reurns as medan coeffcens for posve reurns are close o zero for all observaon perods. For negave reurn frms, by conras, marke reurn does conan consderable nformaon on accruals n almos every counry and he medan coeffcen vares beween 7.6 n perod 1 o 9.9 n perod 4. These 33

34 resuls are n lne wh US resuls of Ball and Shvakumar (2006), alhough our EU resuls show a lower accruals nformaon conen of negave sock reurns on average. 15 The ne effec of ncludng hs addonal marke proxy on he man ACC - CFO coeffcens s repored n panel B of able 7. Coeffcens for posve CFO frms reman farly sable afer ncluson of he marke reurn conrols, wh medan changes varyng beween 1.62 n perod 1 and 1.0 n perod 4. For negave CFO frms, he change n CFO coeffcens vares beween 8.7 n perod 1 and 3.1% n perod 4. Ths s conssen wh negave sock reurns absorbng accruals nformaon from he book-based proxes. Fnally, we fnd ha he rank correlaon of he counry and perod-specfc CFO coeffcens n he book-based regresson specfcaon (2) and he regresson wh addonal reurn conrols s above 0.83 n all observaon perods and close o 1.00 n mos perods. Taken ogeher, hese resuls furher confrm ha addonally ncludng marke-based proxes for economc losses n he orgnal ACC-CFO regressons does no aler nor sgnfcanly mprove he resuls n Table 4. Therefore, he use of he ACC CFO regresson specfcaon as used above s jusfed. 5.4 Influence of Insuonal Counry Facors on Reporng Incenve Effecs Fnally, we nvesgae he nfluence of hree nsuonal counry facors on he reporng ncenve effecs repored n Table 4 (.e., +, 0 and sgns n able 4): () developmen of he equy sock marke (rao of sock marke capalzaon o GDP), () he nfluence of bank deb (rao of bank deb of he prvae bankng secor o GDP), and () unon membershp n 15 Ball and Shvakumar (2006) repor a coeffcen on DRET*RET varyng beween 10% and % dependng on he regresson specfcaon hey use. Ball and Shvakumar sudy a somewha smlar me perod ( ) bu use accruals and cash flow daa from cash flow saemens nsead of relyng on esmaes. 34

35 a counry (obaned from Vsser 2006). Smlar o he logc n Dodge e al. (2007), we argue ha he effec of he frm ncenves (marke pressure, leverage, and labor pressure) on he ACC-CFO relaon mgh vary wh he counres nsuonal facors. For example, labor nensy mgh only affec accruals measuremen o he exen ha labor unons are mporan n a gven counry. Hgh unon membershp mgh herefore nensfy he reporng ncenve snce earnngs numbers are more lkely o be used n wage or resrucurng negoaons. Smlarly, capal marke developmen and he mporance of domesc bank deb n a counry mgh nfluence he effec of marke pressure (resp. leverage) on accruals measuremen. In Table 8 we compue he Spearman rank correlaon beween he counres nsuonal facors and he reporng ncenve effec of marke pressure (SIZE), fnancal leverage (LEVERAGE) and labor nensy (LABOR INTENS) as repored n Table 4, conrollng for me perod and posve versus negave CFO. In he early 1990s, mely gan recognon s more prevalen n counres wh greaer capal marke developmen (as measured by marke cap/gdp). Over me, capal marke developmen becomes less mporan n explanng he mely recognon of gans. Ths fndng s conssen wh he convergence of equy marke developmen n he EU as depced n Fgure 1. The smoohng of posve CFO n hgh leverage frms s nensfed n counres wh greaer domesc deb markes, bu hs correlaon becomes zero n The smoohng of losses ( - sgn n Table 4 panel B) n he economc boom perod s more promnen n counres wh lower deb marke developmen. Conversely, mely loss recognon ( + sgn n Table 4 panel B) n he recesson perod s more pronounced n counres wh smaller deb markes. The correlaon beween he mporance of deb markes and he leverage reporng ncenve effec s close o zero n

36 Fnally, he smoohng of posve CFO s nensfed n counres wh greaer labor unon membershp, especally n he boom perod In he recesson perod , counres wh greaer mely loss recognon seem o be hose wh greaer unon membershp (refleced n he 0.33 posve correlaon). Boh fndngs are n lne wh labor reporng ncenves o be more pronounced n counres where labor relaonshps hsorcally grew as mporan facors of a counry s nsuonal seng. We conclude from hs ha counry nsuonal feaures nensfy he observed effec of frm reporng ncenves for (un)mely loss/gans recognon, and ha busness cycles affec he overall reporng oucome. 6. Concluson In hs sudy, we examne he underlyng deermnans of earnngs comparably n an nernaonal conex. The jon FASB and IASB Concepual Framework emphaszes he mporance of earnngs comparably as a key characersc of decson-useful fnancal nformaon, enablng users o denfy smlares n and dfferences beween underlyng frm economcs. We use a sample of EU counres from 1990 up o he mandaory nroducon of Inernaonal Fnancal Reporng Sandards (IFRS) n The curren seng s parcularly neresng, because Europe wen hrough a perod of hgh economc urbulence, n an effor o mplemen he 1992 Sngle Marke Program o reduce naonal marke enry barrers. Moreover, Europe experenced a mandaed swch o one common se of accounng sandards (IFRS) afer several aemps were made o sandardze naonal GAAP va he nroducon of he 4 h and 7 h Drecve. 36

37 We focus on he accruals accounng sysem and nvesgae he assocaon beween accruals and posve, respecvely negave cash flows across he EU counres. Moreover, we examne he sensvy of hs assocaon o frm-specfc reporng ncenves n each counry whle conrollng for busness cycle. We exend he ACC - CFO regresson model as developed n Ball and Shvakumar (2005) by ncludng reporng ncenves ha are relaed o capal marke, deb fnancng, and labor relaon pressures. Overall, our resuls sugges ha he accrual model s convergng n he EU over a 15 year perod, long before he mandaory mplemenaon of IFRS. In addon, we fnd ha reporng ncenve effecs relaed o equy capal marke, deb marke and labor pressure are key drvers of he underlyng reporng behavor across all observaon perods and connue o exs n The nensy of hese effecs on accrual behavor vares wh busness cycle (recesson versus economc boom perod) and wh nsuonal counry feaures (mporance of sock marke, bank deb and unon membershp). The effec of hese nsuonal feaures seem o dmnsh n 2005, possbly due o he recen elaborae EU naves o enhance corporae governance and enforcemen. The role of busness cycles on reporng behavor n an nernaonal conex s also no well suded before. Our research fndngs are n lne wh earler fndngs as n Ball e al. (2003) and sugges ha a mere swch o a common se of accounng sandards (IFRS) s no necessarly suffcen o brng abou he ancpaed mprovemen n earnngs comparably whn Europe. Our resuls are robus o he ncluson of growh effecs and conrols for a marke-based proxy of economc losses. The curren resuls provde valuable nsghs n he naure of earnngs as a key performance measure and s comparave properes n an nernaonal seng. Moreover, our resuls have 37

38 mporan mplcaons for sandard seers, praconers, he nernaonal nvesmen communy as well as any oher sakeholder confroned wh nernaonal fnancal reporng. Sandard seers have o be aware ha herage of naonal nfluences creaes dfferen ncenves across counres. Ths nrnscally smulaes varaon n fnancal reporng pracces across counres. One way o overcome hs ssue could be o smulae equal enforcemen of reporng qualy across he EU n a smlar way as s currenly enforced n he US va he Secures and Exchange Commsson. For praconers, our resuls call for more rgorous houghs abou he role of Bg-4 accounans n hs EU accounng harmonzaon sory. Snce Bg-4 accounans have consderable nernaonal experse a her dsposal, her awareness and knowledge of naonal dfferences mgh become an mporan asse n smulang more comparable fnancal reporng whn Europe. The nvesmen communy could learn from our sudy by observng ha even afer he mandaory adopon of IFRS n 2005, earnngs measuremen pracces can sll have consderable varaon due o varyng reporng ncenves. Busness cycles and remanng nsuonal dfferences beween counres seem o drve he nensy of hese reporng ncenves. IFRS per se does no resul n hgher fnancal reporng qualy, as refleced n mely gans and loss recognon. 38

39 References Adjaoué, K., & Danhne, J.-P. (2004). Equy reurns and negraon: Is Europe changng? Oxford Revew of Economc Polcy, 20(4), Baele, L., & Inghelbrech, K. (2006). Srucural versus emporary drvers of counry and ndusry rsk. Unpublshed manuscrp, Ghen Unversy. Ball, R., Kohar, S. P., & Robn, A. (2000). The effec of nernaonal nsuonal facors on properes of accounng earnngs. Journal of Accounng and Economcs, 29, Ball, R., Robn, A., & Wu, J. S. (2003). Incenves versus sandards: Properes of accounng ncome n four Eas Asan counres. Journal of Accounng & Economcs, 36(1-3), Ball, R., & Shvakumar, L. (2006). The role of accruals n asymmercally mely gan and loss recognon. Journal of Accounng Research, 44, Ball, R., Robn, A., & Sadka, G.. (2007). Is fnancal reporng shaped by equy markes or by deb markes? An nernaonal sudy of melness and conservasm. Unpublshed manuscrp. Unversy of Chcago. Barh, M. E., Cram, D. P., & Nelson, K. K. (2001). Accruals and he predcon of fuure cash flows. The Accounng Revew, 76(1), Bekaer, G., Harvey, C., Lundblad, C. & Segel, S. (2007). Global growh opporunes and marke negraon, Journal of Fnance, 62(3) pp Beay, A., Weber, J. and Yu, J Conservasm and Deb, MIT Workng Paper. Bronars, S. G., & Deere, D. R. (1991). The hrea of unonzaon, he use of deb, and he preservaon of shareholder wealh. The Quarerly Journal of Economcs, 106, Burgsahler, D. C., Hal, L., & Leuz, C. (2006). The mporance of reporng ncenves: Earnngs managemen n European prvae and publc frms. The Accounng Revew, 81(5), Crooch, M., Talk abou he Concepual Framework, FASB Repor, Augus Cullnan, C. P., & Blne, D. M. (2003). The effecs of labour on accounng choce n Canada. Canadan Accounng Perspecves, 2(2), Darrough, M., & Ye, J. (2007). The declnng comparably of accounng profably: The role of accounng conservasm. Workng Paper Baruch College. Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Deecng earnngs managemen. The Accounng Revew, 70(2),

40 Dechow, P. M., Kohar, S. P., & Was, R. L. (1998). The relaon beween earnngs and cash flows. Journal of Accounng & Economcs, 25(2), Dechow, P. M. and R. Sknner (2000). Earnngs managemen: reconclng he vews of accounng academcs, praconers and regulaors. Accounng Horzons, vol. (2), pp Degeorge, F., & Maug, E. (2006). Corporae fnance n Europe: A survey. Unpublshed manuscrp, Unversy of Lugano, Swzerland. Dodge C., Karoly G.A. and Sulz R.M. (2007) Why do counres maer so much for corporae governance? Journal of Fnancal Economcs, vol.86, pp Erns&Young. (2006). Observaons on he mplemenaon of IFRS. Unpublshed manuscrp. Fama, E. F., & French, K. R. (2004). New lss: Fundamenals and survval raes. Journal of Fnancal Economcs, 73(2), Gerosk, P., & Gugler, K. (2004). Corporae growh convergence n Europe. Oxford Economc Papers, 56(4), Gner, B., & Rees, W. (2001). On he asymmerc recognon of good and bad news n France, Germany and he Uned Kngdom. Journal of Busness Fnance & Accounng, 28(9/10), Gvoly, D., & Hayn, C. (2000). The changng me-seres properes of earnngs, cash flows and accruals: Has fnancal reporng become more conservave? Journal of Accounng & Economcs, 29, Goddard, J., Tavakol, M., & Wlson, J. O. S. (2005). Deermnans of profably n European manufacurng and servces: Evdence from a dynamc panel model. Appled Fnancal Economcs, 15(18), Hand, J. R. M. (2005). The value relevance of fnancal saemens n he venure capal marke. Accounng Revew, 80, Hawawn, G., Subramanan, V., & Verdn, P. (2003). Is performance drven by ndusry- or frm-specfc facors? A new look a he evdence. Sraegc Managemen Journal, 24(1), Hayn, C. (1995). The nformaon conen of losses. Journal of Accounng & Economcs, 20(2), Hung, M. (2000). Accounng sandards and value relevance of fnancal saemens: An nernaonal analyss. Journal of Accounng and Economcs, 30(3),

41 Jenkns, E. (1999). Fnancal reporng n a global capal marke world, Fnancal Accounng Seres 198-A (June 29), pp Joos, P., & Lang, M. (1994). The effecs of accounng dversy: Evdence from he European Unon. Journal of Accounng Research, 32, p Joos, P., & Plesko, G. A. (2005). Valung loss frms. The Accounng Revew, 80(3), Khurana, I., Marn, X., Perera, R. & Raman, K. (2006). Economc sae-varyng ncenves and asymmerc mely recognon of economc losses. Workng Paper, Unversy of Mssour Columba. Klen, A., & Marquard, C. A. (2006). Fundamenals of accounng losses. The Accounng Revew, 81(1), Krschenheer, M. & Melumad, N. (2002), Can smoohng and bg bah accounng co-exs n equlbrum?, Journal of Accounng Research 40(3), pp Land, J., & Lang, M. H. (2002). Emprcal evdence on he evoluon of nernaonal earnngs. The Accounng Revew, 77(Qualy of Earnngs Conference), Leuz, C., Nanda, D., & Wysock, P. D. (2003). Earnngs managemen and nvesor proecon: An nernaonal comparson. Journal of Fnancal Economcs, 69(3), Lbery, S. & Zmmerman, J. (1986). Labor unon conrac negoaons and accounng choces. The Accounng Revew, 61(4), McGahan, A. M., & Porer, M. E. (1997). How much does ndusry maer, really? Sraegc Managemen Journal, 18(Summer Specal Issue), McNamara, G., Vaaler, P. M., & Devers, C. (2003). Same as ever was: The search for evdence of ncreasng hypercompeon. Sraegc Managemen Journal, 24(3), Myers, J., Myers, L. and D. Sknner (2006). Earnngs momenum and earnngs managemen. Workng Paper, Unversy of Chcago. Raonc, I., McLeay, S., & Asmakopoulos, I. (2004). The melness of ncome recognon by European companes: An analyss of nsuonal and marke complexy. Journal of Busness Fnance & Accounng, 31(1/2), Roe, M., (2003). Polcal deermnans of corporae governance, Oxford Unversy Press. Rumel, R. P. (1991). How much does ndusry maer? Sraegc Managemen Journal, 12, Schmalensee, R. (1985). Do markes dffer much? Amercan Economc Revew, 75,

42 Searle, S. R., Casella, G., & McCulloch, C. E. (1992). Varance componens. New York [ec.]: John Wley. Sloan, R. G. (1996). Do sock prces fully reflec nformaon n accruals and cash flows abou fuure earnngs? Accounng Revew, 71(3), Sree, D. (2002), GAAP 2001 Benchmarkng naonal accounng sandards versus IAS: summary of resuls, Journal of Inernaonal Accounng, Audng & Taxaon, 11, pp Sweeney, A. P. (1994). Deb-covenan volaons and managers' accounng responses. Journal of Accounng & Economcs, 17(3), Van Tendeloo, B. & A. Vansraelen (2005). Earnngs managemen under German GAAP versus IFRS. European Accounng Revew, vol.(1) p Vsser, J. (2006) Unon membershp sascs n 24 counres, Monhly Labor Revew, January, pp Was, R. L., & Zmmerman, J. L. (1986). Posve accounng heory. Englewood Clffs, N.J.: Prence-Hall Inernaonal. Wu, J. & I. Zhang (2007). Volunary IAS and US GAAP adopon by connenal European frms: he role of labour unons. Workng Paper, Unversy of Rocheser. 42

43 Tables and Fgures Table 1: Volunary IFRS adopers n he European Unon per year ( ) (1) Counry SUM AUSTRIA BELGIUM DENMARK FINLAND FRANCE GERMANY GREECE IRELAND ITALY NETHERLANDS PORTUGAL SPAIN SWEDEN UNITED KINGDOM New IFRS adopers Populaon (1) Ths Table repors all yearly observaons of EU-15 (less Luxembourg) non-fnancal frms adopng IFRS for he frs me n her corporae hsory and ha connue reporng under IFRS for (a leas) wo consecuve years afer he volunary adopon year (no requred for mandaory adopers n 2005). We denfed volunary and mandaory IFRS adopers by checkng he Worldscope em WC07036: Acccounng Sandards Followed. We code an accounng sandard as IFRS when WC07536 equals code 02 ( Inernaonal Sandards ), 06 ( Inernaonal sandards and some EU gudelnes ) and 23 (IFRS). We code an accounng sandard as local GAAP when WC07536 s no equal o IFRS (as descrbed before nor s equal o US GAAP: (code 03 (US GAAP), code 13 (US GAAP, nconssency problems) or code 20 (US GAAP reclassfed from local sandards)). Noe ha frms lsed on he Alernave Marke Index (London) are sll allowed o repor accordng o local UK GAAP and ha oher exempons (especally small frms lsed on SME segmens n varous counres) exs.. 43

44 Table 2: Descrpve Sascs (All frms) (1) * [1] [2] [3] [4] Mean earnngs cash from operaons accruals hgh-ech marke cap (sze) , ,538.3 fnancal leverage labor nensy Medan earnngs cash from operaons accruals Hgh-ech marke cap (sze) fnancal leverage labor nensy Sandard devaon Earnngs cash from operaons Accruals hgh-ech marke cap (sze) 1, , , ,449.9 fnancal leverage labor nensy observaons 5,595 7,931 9,536 2,048 (1) The sample consss of 25,110 frm-year observaons from EU counres used n he regressons shown n able 4 (afer ouler deleon). Local currences have been ransformed o euros. Earnngs, cash from operaons and accruals are scaled by oal asses. Earnngs are measured as ne ncome afer preferred dvdends, Accruals are calculaed as (ΔCA - ΔCASH) - (ΔCL - ΔSTDEBT - ΔTAXPAY) DEPR where ΔCA s he change n curren asses (WC02201) from year o -1, ΔCASH s he change n cash (WC02001), ΔCL s he change n curren lables (WC03101), ΔSTDEBT s he change n shor erm deb (WC03051), ΔTAXPAY s he change n axes payable (WC03063) and DEPR s he annual deprecaon and amorzaon expense n year (WC01151), where WCnumber represens he Thomson Fnancal Worldscope code. The las 4 varables represen he conrol varables n he regresson model n able 4: hgh-ech (a dummy varable equal o 1 f he frm belongs o a hgh-ech ndusry (aerospace & defense 271, pharma & boech 457, fxed lne elecom 653, moble elecom 657, sofware & compuer svc 953, echnology hardware & equpmen 957), marke capalzaon (value of equy measured a end of fscal year, expressed n mllon euro), fnancal leverage (shor-erm and long-erm deb dvded by oal asses), and labor nensy (number of employees dvded by marke capalzaon). We use he Indusry Classfcaon Benchmark, also known as FTSE Acuares Classfcaon Sysem, o deermne wheher a frm belongs o a hgh-ech secor. All fnancal and nsurance frms are excluded from he sample. (*) For 2005, only IFRS adopers are ncluded n he able. CFO s he repored CFO (he cash flow saemen s requred under IFRS), and accruals are calculaed as he dfference beween earnngs and CFO. 44

45 Table 3: Descrpve Sascs (By counry) (1) Varable perod AUT BEL DNK FIN FRA GER GRC IRL ITA NLD PRT SPA SWE UK Earnngs [1] [2] [3] [4] CFO [1] [2] [3] [4] Accruals [1] [2] [3] [4] Hgh-ech [1] [2] [3] [4] Marke cap [1] (sze) [2] , , , , , , , [3] , , [4] 1, , , , , , , , , , , ,077.9 Leverage [1] [2] [3] [4] Labor nensy [1] [2] [3] [4] (1) The able shows average values of each varable over 4 perods for each of he EU counres. All varables are defned n Table 1. The perod numbers refer o ([1], conracon perod), ([2], expanson perod), ([3], conracon perod) and 2005 ([4], expanson perod). 45

46 Table 4: Mulvarae Regresson of Accruals on Cash Flows Per Sub-Perod Conrollng for Reporng Incenves and Indusry (1) Panel A: posve CFO NUMBER OF OBSERVATIONS β 1 COEFFICIENT EFFECT ON β 1 WHEN CONTROLLING FOR SIZE LEVERAGE LABOR INTENS [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AUSTRIA * * * * BELGIUM * * * * DENMARK * * * * FINLAND * * * * FRANCE * * * * GERMANY * * * * GREECE * * * * IRELAND * * * ITALY * * * * NETHERL * * * * PORTUGAL * * * * SPAIN * * * * SWEDEN * * * * UK * * * * AVG STD Panel B: negave CFO NUMBER OF OBSERVATIONS β 2 COEFFICIENT EFFECT ON β 2 WHEN CONTROLLING FOR SIZE LEVERAGE LABOR INTENS [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AUSTRIA * BELGIUM * DENMARK * FINLAND * FRANCE * * * GERMANY * * 10.9 * * GREECE * * * * IRELAND ITALY * *.5 * * NETHERL * * PORTUGAL * - * SPAIN * * SWEDEN * 20.2 * 16.5 * UK * 9.3 * AVG STD * 46

47 (1) The able repors he esmaon resuls for he followng regresson: ACC = α = 1 1, = 1 1, = 1 1, α DPCFO + = 1 1, β DPCFO CFO + α = 1 2, = 1 2, γ DPCFO INCENTV + DNCFO = 1 2, δ DPCFO CFO INCENTV + β DNCFO CFO γ DNCFO INCENTV = 1 2, δ DNCFO CFO INCENTV + ε The regresson s separaely esmaed for four separae sub perods, [1]: (recesson/conracon), [2]: (expanson), [3]: (recesson/conracon) and [4]: 2005 (expanson and frs IFRS adopon year). We exclude he op and boom 1% oulers based on a mulvarae deleon rule for observaons havng he larges ouler effecs on our regresson resuls. Resuls reman sable f we remove he op and boom 0.5% (2%). Ths able ncludes all counry-frm observaons and boh posve and negave CFO (panel A and B respecvely). The columns wh he β 1 (respecvely β 2 ) COEFFICIENT esmaes (mulpled by 100) shows he regressons esmaes on he counry s CFO for a resrced model,.e. whou ncludng any conrol varable ( γ 1, γ 2, δ1, δ 2 = 0). The asersk ndcaes wheher he coeffcen s sgnfcan a he 5% level. The number of observaons refers o he observaons for a specfc counry n one of he four me perods. The number of observaons n perod [4] s less han one hrd of he oal number of observaons n he prevous perods manly because 2005 frm-daa are only ncluded f he frm repors accordng o IFRS sandards. Specfc marke segmens (lke he AIM segmen n London) allow frms o repor n accordance wh local GAAP unl In he rgh hand sde of Table 4, we nclude resuls from esmang he full model ( γ 1, γ 2, δ1, δ 2 0), where he varable INCENTV s respecvely he sandardzed (bounded beween 0 and 1) frm-rank sze (marke capalzaon), leverage (shor-erm and long-erm deb scaled by oal asses), hgh-ech (a dummy varable equal o 1 f he frm belongs o a hgh-ech ndusry, as defned n Table 1), and labor nensy (number of employees scaled by marke capalzaon) whn a counry. Repored n he Table 4 s he mpac of δ 1 on β 1 (respecvely δ2 on β2 n panel B). In parcular, a + n he able ndcaes a sgnfcan (a 5% level) effec of he INCENTV varable on he counry-cfo β 1 (respecvely β 2 ) coeffcen, ndcaes a sgnfcanly negave effec, and 0 ndcaes no sgnfcan effec. A nex o he ncenve conrol effec ndcaes ha he INCENTV effec absorbs he man counry-cfo β 1 (respecvely β 2 ) effec. 47

48 Table 5: Hgh-Tech Indusry Membershp Effec on Accruals CFO correlaons (1) Panel A: posve CFO β 1 COEFFICIENT FROM RESTRICTED MODEL β 1 (low ech) FROM FULL MODEL δ (hgh ech effec) 1 FROM FULL MODEL [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AVG MED STD STD/AVG Panel B: negave CFO β 2 COEFFICIENT FROM RESTRICTED MODEL β 2 (low ech) FROM FULL MODEL δ (hgh ech effec) 2 FROM FULL MODEL [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AVG MED STD STD/AVG (1) Ths able repors he esmaon resuls for: ACC = α 0 + α DPCFO + DNCFO + DPCFO xcfo = = = + 1 1, α 1 2, β 1 1, β = 1 2, DNCFO xcfo DPCFO xhitech + γ DNCFO xhitech + δ DPCFO xcfo xhitech + δ = 1 1, = 1 2, = 1 1, = 1 2, γ DNCFO xcfo xhitech + ε The regresson s esmaed separaely for four separae sub perods, [1]: (recesson/conracon), [2]: (expanson), [3]: (recesson/conracon) and [4]: 2005 (expanson and frs IFRS adopon year). We exclude he op and boom 1% oulers based on a mulvarae deleon rule for observaons havng he larges ouler effecs on our regresson resuls. Resuls reman sable f we remove he op and boom 0.5% (2%). Ths able ncludes all counry-frm observaons and boh posve and negave CFO (panel A and B respecvely). The resrced model resrcs he γ 1, γ 2, δ 1, and δ 2 o zero, and corresponds wh he resuls from able 4. 48

49 Table 6: Growh Effecs on Accruals CFO correlaons (1) Panel A: posve CFO β 1 COEFFICIENT FROM RESTRICTED MODEL δ 1 xgrowth β 1 + neg FROM FULL MODEL δ 1 xgrowth β 1 + pos FROM FULL MODEL [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AVG MED STD STD/AVG Panel B: negave CFO β 2 COEFFICIENT FROM RESTRICTED MODEL δ 2 xgrowth β 2 + neg FROM FULL MODEL δ 2 xgrowth β 2 + pos FROM FULL MODEL [1] [2] [3] [4] [1] [2] [3] [4] [1] [2] [3] [4] AVG MED STD STD/AVG (1) Ths able repors he esmaon resuls for: ACC = α 0 + α DPCFO + DNCFO + DPCFO xcfo = = = + 1 1, α 1 2, β 1 1, β = 1 2, DNCFO xcfo DPCFO xgrowth + γ DNCFO xgrowth + δ DPCFO xcfo xgrowth + δ = 1 1, = 1 2, = 1 1, = 1 2, γ DNCFO xcfo xgrowth + ε 49

50 Ths regresson dffers from regresson (2) n ha we addonally nclude a GROWTH varable, defned as he sandardzed whn-counry frm-rank of asse growh. The able repors summary sascs (mean, medan, sandard devaon) of he counry regresson esmaes for posve (panel A) and negave cash flow observaons (panel B) separaely. The frs four columns repor he average β 1 (respecvely β 2 ) coeffcens when no growh conrol s ncluded n he model (.e. when γ 1, γ 2, δ1, δ 2 = 0 ; equal o he resuls n β 1 (respecvely β 2 ) -columns of Table 4). Columns 5 o 8 repor he average sum of he respecve β 2, coeffcen plus δ 2, negave growh observaons. The las four columns conan smlar resuls bu for posve growh observaons ( GROWTH pos, represens he average growh rank for posve growh observaons. mes he average growh rank for β + ) where 1, δ1, x GROWTH pos, 50

51 Table 7: Marke Reurn as an Addonal Proxy for Gans and Losses (1) Panel A β 3 (%) 3 β4 β + (%) [1] [2] [3] [4] [1] [2] [3] [4] MEDIAN MIN MAX STDEV Panel B β1 -change (%) β 2 -change (%) [1] [2] [3] [4] [1] [2] [3] [4] MEDIAN MIN MAX STDEV RANKCORR (1) The able repors summary esmaon resuls for he followng regresson: ACC = α + α DPCFO α 2 DNCFO α 3 DRET 0 = 1, = 1, = 1, β = 1 1, DPCFO xcfo + DNCFO xcfo + β RET + β = 1 2, = 1 3, = 1 4, β DRET xret + ε Ths regresson dffers from regresson (2) n ha we here addonally nclude sock reurns as markebased proxes for economc gans (RET) and losses (DRET*RET), by addng () a dummy varable for negave sock reurns (DRET), () sock reurns (RET) and () he neracon effec for negave reurns (DRET*RET) o he nal regresson model. Sock reurns are measured as he reurn on frm from 9 monhs before fscal year-end o hree monhs afer fscal year-end as n Basu (1997). Panel (A) repors summary sascs (mean, medan, mnmum, maxmum and sdev) of he counry RET (lef column) and DRET (rgh column) coeffcens. Panel (B) repors summary sascs of he ncrease (+) versus decrease ( ) for he β1 and β 2 coeffcens when he full model above s compared o a more resrced model where no marke-based reurn conrols are ncluded (.e., α 3, β3 and β 4 =0) and hs boh for posve (lef column) and negave (rgh column) CFO coeffcens. In addon, he rank correlaon of counry-ranked regresson coeffcens before and afer ncludng marke-based reurn conrols. 51

52 Table 8: Correlaon beween Insuonal Counry Facors and Frm Incenve Effecs on he ACC-CFO Correlaon n Table 4 (1) Insuonal Facor Perod Reporng Incenve Effec SIZE effec (pos CFO) δ1 SIZE poscfo / β 1 SIZE effec (neg CFO) δ 2 SIZEnegCFO / β 2 marke cap / GDP [1] [2] [3] [4] LEVERAGE effec (pos CFO) δ1 LEV poscfo / β 1 LEVERAGE effec (neg CFO) δ 2 LEVnegCFO / β 2 dom cred / GDP [1] [2] [3] [4] LABOR INTENS (pos CFO) δ1 LABOR poscfo / β 1 LABOR INTENS (pos CFO) δ 2 LABORnegCFO / β 2 unonzaon [1] [2] [3] [4] (1) The able repors he Spearman rank correlaon beween hree nsuonal facors measured a he counry level and he respecve reporng effecs on he ACC-CFO coeffcens n able 4,.e., δ1 INCENTV poscfo / β for posve CFO frms measured a 1 he counry level and δ 2 INCENTVnegCFO / β for negave CFO frms. Conssen wh 2 able 4, he Spearman correlaon coeffcens are compued across he EU counres for each subperod n order o conrol for he busness cycle effec. The mporance of publc equy (marke cap/gdp), and domesc bank deb (domesc cred/gdp) are obaned from he Worldbank. The degree of unon membershp (unonzaon) measures he srengh of labor unons for each of he EU counres over he perod and s obaned from Vsser (2006). 52

53 Fgure 1: Macro-Economc Convergence Paerns (1) 100% 90% CPI GDP/cap UNEMPLOY MCAP/GDP DEBT/TA FOREIGNSALES 80% 70% 60% 50% 40% 30% 20% 10% 0% (1) Ths Fgure dsplays he yearly dsperson of macro-economc and nsuonal varables across EU counres and s measured as: (sandard devaon of varable /average of varable). CPI measures nflaon raes and refers o he Consumer Prce Index, GDP/cap o he Gross Domesc Produc per capa and Unemploy s he unemploymen dsperson across EU counres. MCAP/GDP s he marke capalzaon of all lsed frms n ha counry dvded by he Gross Domesc Produc, Deb/TA s he rao of frm deb o oal frm asses and Foregnsales s he proporon of foregn sales o oal sales. CPI, GDP/capa, unemploymen daa and MCAP/capa come from Worldbank records, whle Deb/TA and Foregnsales are aken from Worldscope. 53

54 Fgure 2: CEPR Recesson Expanson ndcaor n Europe (1) (1) Ths graph shows he values of he EuroCOIN,.e. he monhly ndcaor of he euro area busness cycle publshed by CEPR for he perod January 1988 o February The shaded areas n he char correspond o perods whch he ndcaor suggess are cyclcal expansons. 54