Market timing and capital structure: Evidence from a decomposition of the market-to-book ratio. Salma Kasbi 1. Université Paris-Dauphine, DRM-CEREG

Transcription

1 Marke iming and capial srucure: Evidence from a decomposiion of he marke-o-book raio Salma Kasbi 1 Universié Paris-Dauphine, DRM-CEREG This sudy examines he impac of marke iming on capial srucure. Using a decomposiion of he marke-obook raio ino misvaluaion and growh opporuniies, developed in Rhodes-Kropf, Robinson and Viswanahan (2005), I find ha he negaive relaionship beween he hisorical marke-o-book raio and leverage, documened in Baker and Wurgler (2002) canno be only aribued o growh opporuniies. I also reflecs he persisen impac of pas iming aemps. EFM Classificaion : 140 Keywords: Marke Timing, Capial Sucure 1 Cereg-DRM, Universié Paris-Dauphine, place du Maréchal de Lare, Paris cedex salma.kasbi@gmail.com. 1

2 1. Inroducion Three major heories offer conflicing predicions abou he deerminans of firms capial srucures. The radeoff heory saes ha firms adjus heir capial srucure over ime, oward an opimal leverage resuling from balancing he coss and he benefis of deb financing (e.g. Jensen and Meckling (1976), Myers (1977), Sulz (1990)). The pecking order heory (Myers and Majluf (1984), Myers (1984)) is based on he exisence of informaion asymmery beween managers and ouside invesors ha cause exernal finance o be cosly. This heory saes ha he relaive coss of inernal and exernal finance are he major deerminans of firms capial srucures. As he coss of inernal financing are lower han he coss of exernal financing, firms will prefer inernal funds. When exernal financing is needed, firms will prefer deb hen equiy as a las resor, o mee heir financial deficis. A recen srand of he corporae finance lieraure focuses on he marke iming heory. This heory posis ha using heir superior insider informaion, managers are able o ake advanage of windows of opporuniies o successfully ime heir equiy offerings and ha he iming of pas securiies issuance is a major deerminan of curren capial srucures. In heir prominen sudy, Baker and Wurgler (2002) asser ha capial srucure is he cumulaive oucome of pas aemps o ime he equiy marke. They consruc he exernal finance weighed average marke-o-book (henceforh M/B efwa ) o capure he marke iming aemps: his variable akes high values when a firm raises exernal finance (equiy or deb) a 2

3 imes where is marke valuaion is high and lower values oherwise. As hey find he M/B efwa o be negaively correlaed wih he leverage raio, he auhors argue ha he effec of marke iming is very persisen and ha firms do no readjus heir capial srucure owards a arge. Their resuls are difficul o reconcile wih he radiional heories of capial srucure. The marke iming heory, and paricularly he empirical findings of Baker and Wurgler (2002) which are he focus of his paper, generaed a heaed debae. Evidence on marke iming is suppored by he Graham and Harvey s (2001) survey where sock mispricing appear o be an imporan or very imporan consideraion in he decision of issuing equiy for abou wo-hirds of CFOs. This observaion is in line wih a significan number of empirical sudies ha documen ha firms ime heir equiy offerings and end o issue equiy following a sock price appreciaion. Among ohers, Loughran and Rier (1995), Pagano, Panea, and Zingales (1998) show ha firms end o underake IPOs when heir indusry marke valuaions are high. Marsh (1982), Lucas and McDonald (1990), Jung, Kim, and Sulz (1996) and Hovakimian, Opler and Timan (2001) find ha seasoned equiy offerings are srongly relaed o sock prices. Following Baker and Wurgler (2002) empirical resuls, academic ineres in he impac of he iming of securiies issuance on curren capial srucure has gained momenum. Despie widespread agreemen on he emporary effec of marke iming on capial srucure, he long-lasing impac of his phenomenon remains very conroversial and a number of recen sudies challenge Baker and Wurgler (2002) findings. Leary and Robers (2005) argue ha firms acively rebalance heir leverage so ha he impac of marke iming vanishes wihin hree o five years following equiy issuances. They conclude ha firms follow a dynamic radeoff. Ali (2006) analyses he impac of iniiaing an IPO in a ho or a cold marke, as a measure of marke iming. He shows ha he iniial impac of ho issues 3

4 markes on leverage is consequenly balanced away whihin a wo-years period. Flannery and Rangan (2006) and Kayhan and Timan (2007), also find ha he impac of marke iming is shor-lived. Conversely, Fama and French (2002) and Welch (2004) provide evidence on a slow adjusmen speed. More recenly, Huang and Rier (2007), using a ime varying equiy risk premium o capure he iming aemps, provide evidence on he persisen impac of Marke iming on capial srucure. They argue ha firms adjus very slowly owards heir arge leverage. The second conroversy surrounding he marke iming heory quesions he relevance of he use of he hisorical marke-o-book raio o appropriaely proxy for a firm s marke iming aemps. Several papers (e.g. Kayhan and Timan (2007), Hovakimian (2006)) aes ha he observed negaive relaionship beween MB efwa and leverage arises because he MB efwa conains informaion abou growh opporuniies ha is no adequaely capured in curren marke-o-book. In paricular, he marke-o-book raio is widely used as a proxy for firms growh opions, which should be negaively relaed o leverage. The pecking order heory predics ha firms wih higher marke o book raios, which are likely o have higher growh opporuniies, may issue equiy o finance heir curren se of invesmens when hey run ou of heir inernal funds and of heir deb capaciy. According o he radeoff heory, firms wih higher growh opporuniies keep lower leverage levels o keep heir financial flexibiliy o avoid he underinvesmen problem (Myers, 1977). Therefore, o properly invesigae he impac of marke iming on capial srucure, i is crucial o isolae misvaluaion from he oher componens of marke valuaion such as he growh prospecs. In his paper, I es he marke iming heory using a more accurae measure of firms misvaluaions han he marke-o-book raio, ha is no conaminaed by growh opporuniies. Following Rhodes-Kropf, Robinson and Viswanahan (2005, henceforh 4

5 RKRV), I breakdown he marke-o-book raio ino hree componens: firm-specific error, ime-series secor error and long-run value-o-book. RKRV measure firm-specific error as firm-specific deviaions from valuaions implied by conemporaneous secor muliples. This variable capures a firm s idyosyncraic misvaluaion. Time-series secor error reflec he deviaion of curren secor muliples from long-erm secor muliples. This componen is mean o capure he misvaluaion of a secor. The las componen of he marke-o-book which is he long-run value-o-book, measures he discrepancy beween long-run secor muliples and book value. This componen is inended o accoun for growh opporuniies. I subsequenly compue he exernal finance weighed-averages of hese hree componens o es wheher he negaive relaionship beween M/B efwa and curren capial srucure documened in Baker and Wurgler (2002) arises because of he persisence of he marke iming s impac or is simply driven by growh opporuniies, as implied by radiional heories. Overall, my empirical findings srongly suppor he marke iming heory. Once he effecs of growh opporuniies isolaed, he remaining componen of M/B efwa which reflecs misvaluaion shows a negaive and persisen impac on leverage. Moreover, I provide evidence on he abiliy of he iming measure I use o properly capure misvaluaion. This variable is unrelaed o invesmen bu posiively associaed wih he firm s level of cash. These resuls are in opposiion wih he predicions of he radeoff heory and he pecking order heory. Furhermore, he resuls suppor he idea ha firms ime heir equiy issues bu also heir deb issues. An equiy issuance is preceded by a significan runup of he firm s specific-misvaluaion componen and is followed by a significan decline of his variable. Conversely, firms experience a decline of heir overvaluaion componen before a deb issuance, immediaely followed by a significan upward rend. These paerns are observable 5

6 hree years before and hree years afer a securiy issuance. While sudies using he marke-obook raio as a iming measure widely documened he former phenomenon, hey failed in providing evidence on he iming of deb issues (Hovakimian (2006)). Finally, I find ha equiy issues are also imed o periods of high secor valuaions. However, hose marke imers are also more likely o have negaive newly reained earnings and hey subsequenly issue relaively more deb so ha he iniial impac of equiy issues on leverage is offse. This resul is puzzling and consisen wih he conclusions of Ali (2006) regarding ho-markes firms. The remainder of his paper proceeds as follows: Secion 2 provides he mehodology, Secion 3 presens daa, Secion 4 presens he empirical resuls and Secion 5 concludes. 2. Mehodology 2.1. Previous mispricing measures Baker and Wurgler s (2002) marke iming heory saes ha a firm s capial srucure is he cumulaive oucome of pas aemps o ime he equiy marke. Their resuls show ha he MB efwa is negaively relaed o curren leverage and consequenly, ha low leveraged firms are hose who raised equiy when heir valuaions were high, as measured by he marke-obook raio, and vice versa for high leveraged firms. Subsequen empirical ess of he marke iming heory (e.g. Hovakimian (2006) and Kayhan and Timan (2007)) demonsraed ha he use of he marke-o-book raio o infer mispricing can be misleading. As a maer of fac, radiional heories commonly view he marke-o-book raio as a measure of a firm s se of invesmen opporuniies, which when high, leads he firm o increase is equiy issuances. Several papers have used oher mehods o es for he impac of misvaluaion on capial srucure. A recen paper of Ali (2006), invesigaes he impac of iming aemps on changes 6

7 in leverage by idenifying as make imers hose firms ha wen public on ho markes. They documen a shor-living marke iming impac. In paricular, hey find ha ho marke firms increase heir leverage in he wo years following heir IPOs by issuing significanly more deb and less equiy han cold-marke firms. However such a es doesn direcly focus upon he misvaluaion of he firm. If i capures misvaluaion, i would be a bes, some par of a firm s equiy misvaluaion ha is shared by he whole marke. Furhermore, his es is unable o quanify he exen of mispricing. Huang and Rier (2007) poin ou ha because of he IPO marke cycles, i is imporan o conrol for he difference in marke condiions and firm characerisics beween ho and cold IPO firms. Oher approaches use variables such as insider rading or analys coverage o infer marke iming. Ye, hese measures are also noisy marke iming proxies. This paper uses a more direc measure of a firm s mispricing, based on he mehodology developped by RKRV 2, ha idenifies sock misvaluaion by filering growh opporuniies Marke-o-book raio decomposiion RKRV decompose he marke-o-book raio ino marke-o-value and value-o-book. While hey use he marke-o-book raio for equiy, his paper requires he decomposiion of he marke-o-book raio of asses for purposes of accordance wih Baker and Wurgler (2002) specificaion. M/B M/V x V/B (1) 2 RKRV (2005) implemen a decomposiion of he marke-o-book raio o sudy how valuaion waves affec merger waves. 7

8 where, M is he firm s marke value of asses, V is he rue value of asses and B is he book value of asses. Marke value is defined as he sum of equiies marke value and book deb. Expressing Eq. (1) in logarihms gives : m b (m v) + (v b) (2) where m, v and b denoe he logarihms of M, V and B, respecively. The firs componen of Eq.(2) is inended o capure a firm s poenial misvaluaion. If markes perfecly anicipae he rue value of he firm (by knowing is fuure growh opporuniies, discoun raes and cash flows), hen here is no room for misvaluaion and he firs componen will be equal o zero. Oherwise, if he markes overvalue (undervalue) he firm s rue value, hen (m v) would be posiive (negaive). The second erm of Eq.(2) capures he divergence beween a firm s rue value and he curren book value of is asses. Thus, his erm capures a firm s growh prospecs. Furhermore, for a given firm, he misvaluaion par of he marke-o-book raio incorporaes a firm-specific componen and a second misvaluaion componen, ha is common o all firms of he same secor or marke. RKRV express he rue value v as a linear funcion of firm-specific accouning informaion a ime, θ i, and a vecor of condiional accouning muliples, α. The hree componens of he marke-o-book are hus esimaed for a firm i, in a secor j, a ime, using his equaion: m i b i m v( ; ) v( ; ) v( ; ) v( ; ) i i j i j i j i j i firmspecificerror ime series secor error longrun valueo book b (3) where, ime accouning muliples are represened by α j, while long-run muliples are represened by α j. 8

9 The firs erm (m i v(θ i ;α j ))measures he difference beween he firm s marke value and is fundamenal value as implied by is accouning muliples θ i and is secor j muliples α j measured a he valuaion year. This erm represens he firm-specific error. The second erm (v(θ i ;α j ) v(θ i ;α j )) measures he difference beween he firm s fundamenal value condiional on conemporaneous accouning muliples and is value implied by is accouning informaion and long-run muliples. This erm, called ime-series secor error, reflecs he whole secor misvaluaion a ime. The hird componen (v(θ i ;α j ) b i ) measures he difference beween he firm s valuaion based on long-run muliples and is book value. This las componen capures he firm s se of invesmen prospecs a ime. RKRV s mehodology 3 for esimaing v, is based on a popular valuaion echnique widely used in he accouning lieraure : he residual income model (Ohlson 1995). This model defines he firm s rue value as he presen value of he fuure abnormal earnings of he firm plus is book value. M B E ( ROA r ) B (1 r ) 1 (4) Where ROA is he reurn on asses and r he cos of capial. The residual income model can be implemened using analys earnings forecass bu his mehod can be problemaic as analys forecass are likely o be biased oward more growh opporuniies, as poined ou by RKRV and Rier and Warr (2002). 3 For deails on he mehod, cf. RKRV (2005). As I discussed earlier, he difference beween he original mehodology ha RKRV (2005) develop and he one in his paper is due o he use of he marke and book values of asses insead of he marke and book values equiy. 9

10 M Following RKRV, I esimae he marke value of asses using he equaion below: B NI (5) Esimaing Eq. (5) requires some assumpions. The expeced fuure ROA is a consan muliple of expeced fuure discoun raes, book asses and ne income are growing a consan raes. The ne income allows for capuring he value of inangibles ha no conained in he book value of asses. Expressing Eq. (5) in logarihms yields o: m i b Ln NI I LnNI (6) 0 j 1 j i 2 j i 3 j ( 0) i i Where ln(ni)+ is he absolue value of ne income and I<0 is an indicaor funcion for observaions wih negaive ne income. This allows for aking ino accoun he observaions wih negaive ne income, while esimaing he equaion above using logarihms. To esimae Eq. (6) above, I run annual, cross-secional regressions on firms grouped according o he 11 Fama and French indusries (financial firms, ha are firms of he 11 h Fama and French indury are no included). The esimaed value of v(θ i ;α j ) for a firm i, in a secor j, a ime is given by he fied values from Eq. (6) above: v( b, ; ˆ, ˆ, ˆ, ˆ ) ˆ ˆ ˆ i NI i 0 j 1 j 2 j 3 j 0 j 1 j bi 2 j Ln 3 j ( 0) NI ˆ I LnNI (7) i i To compue he long-run secor muliples, I average over ime, for each indusry, he esimaed values of he conemporaneous muliples : 1 T kj kj, k = 0, 1, 2, 3. v( bi, NI i; 0 j, 1 j, 2 j, 3 j ) 0 j 1 j bi 2 j Ln 3 j ( 0) NI I LnNI (8) i i 10

11 3. Daa 3.1. Sample and variables consrucion The iniial sample comprises all firms on COMPUSTAT during he period I resric he sample o exclude financial firms wih a Fama and French indusry code 11, and observaions for which variables ha allow for he calculaion of he hree componens of he marke-o-book raio 4 are lacking. The number of remaining observaions in his sample is 103,600. This large sample allows for a more precise calculaion of he previously menioned α muliples. Table 1 displays he yearly number of firms and he number of firms for each indusry, used o compue he marke-o-book componens. I furher form wo samples: an IPO-ime sample and a calendar ime sample. The calendar ime sample is iniially formed by he firms of he sample described above. The IPO-ime sample is he sub-se of he calendar-ime sample ha consiss of all firms for which I could deermine an IPO dae from he SDC daabase. The sample is hen resriced o firms ha iniiaed heir IPO beween 1986 and Firms are dropped from he SDC sample if hey have more han one IPO dae repored. The IPO year, is he fiscal year during which he IPO akes place. Following Baker and Wurgler (2002), firms wih a minimum book value of asses below $10 million, and firms wihou complee daa on oal asses beween he IPO year (or he firs year he firm enered COMPUSTAT for he calendar-ime sample) and he year he firm exis COMPUSTAT are excluded from he samples. Firm-year observaion ouliers are dropped according o some crieria described below. Book equiy (E) is defined as oal asse (daa iem 6) minus oal liabiliies (daa iem 181) and preferred sock (daa iem 10) plus 4 Observaions wih missing or negaive daa on book asses (daa iem 6), price (daa iem 199), and common shares ousanding (daa iem 25) or missing daa on ne income (daa iem 172). 11

12 deferred axes (daa iem 35) and converible deb (daa iem 79). If preferred sock is missing i is replaced wih he redempion value of preferred sock (daa iem 56). Book deb (D) is defined as oal asses minus book equiy. Book leverage (D/A) is defined as book deb divided by oal asses. Marke equiy is defined as he share price (daa iem 199) imes common shares ousanding (daa iem 25). The marke-o-book raio (M/B) is defined as he marke value of asses divided by oal asses. The Marke value of asses is he sum of oal asses minus book equiy plus marke equiy. Ne equiy issuance (e/a) is he change in book equiy minus he change in balance shee reained earnings (daa iem 36) divided by asses. Newly reained earnings ( RE/A) are defined as he change in reained earnings divided by asses. Ne deb issuance (d/a) are defined as he residual change in asses divided by asses. Profiabiliy (EBITDA/A) is defined as earnings before ineres, axes and depreciaion (daa iem 13) divided by asses. Firm size (SIZE) is he logarihm of ne sales (daa iem 12). Tangibiliy (PPE/A) is defined as ne plan, propery and equipmen (daa iem 8) divided by asses. Capial expendiures (CAPEX/A) (daa iem 128) are scaled by oal asses. Research and developmen expenses (RD/A) (daa iem 46) are divided by asses. CASH/A is he raio of cash and shor-erm invesmens (daa iem 1) divided by asses. Observaions are dropped if M/B is above 10. Observaions wih book leverage ha exceeds 1 are also excluded. All he scaled variables, defined above, are by fiscal year end oal asses Marke iming measures This secion describes he consrucion of he previously oulined measures of marke iming opporuniy in more deail. Baker and Wurgler (2002) inroduced he exernal finance weighed-average marke-obook raio as a proxy for marke iming. They define i as follows: 12

13 M/ B e d is is i M/ B is s0 eir dir r0 (9) Where, 0 is he IPO dae or he firs year he firm enered COMPUSTAT. e and d denoe he amoun of ne equiy and ne deb issued, respecively. Following Baker and Wurgler (2002), negaive amouns of exernal finance are rese o zero. As in Baker and Wurgler (2002), I use he exernal finance weighed-average. Bu as I decomposed he marke-o-book, he weighing is based on each of he hree componens insead of he marke-o-book iself. For compuing purposes, he averaged variable mus be posiive. One way for geing posiive variables is o use heir exponenial form. This also allows for having raios similarly o an analysis based on he marke-o- book raio. Therefore he hree exernal finance weighed averages are defined as follows: o Exernal finance weighed average firm-specific misvaluaion: Fmis e d is is i Fm is is s0 eir dir r0 (10) o Exernal finance weighed average secor misvaluaion: Smis e d is is i Sm is is s0 eir dir r0 (11) o Exernal finance weighed average growh opporuniies: Growh e d is is i Growh is s0 eir dir r0 (12) 13

14 Fmis i is he raio of he firm s marke value of asses divided by is esimaed rue value condiional on secor conemporaneous muliples. Fmis i is he exponenial of he esimaed firm-specific error. Fmis i Exp i m v ; i j M i V ; i j (13) Smis i is he raio of he firm s esimaed conemporaneous fundamenal value of asses divided by is esimaed long-run fundamenal value. Smis i is he exponenial of he esimaed ime-series secor-error. Smis i Exp v ; v ; i j i j V i; j V ; i j (14) Growh i is he raio of he firm s long-run fundamenal value of asses divided by he book value of asses. Growh i is he exponenial of he esimaed long-run error. Growh i Exp i j v ; b i V i; j B i (15) Observaions where Fmis, Smis, Growh, Fmis efwa, Smis efwa, Growh efwa and M/B efwa are above 10 are excluded from he samples. The M/B efwa is he weighed average of pas marke-o-book raios. The weigh for each year is he amoun of exernal finance raised in ha year relaive o he oal exernal finance raised by he firm since he IPO year. Subsequenly, firms ha issue securiies when heir valuaions are high will end o have high values of M/B efwa. Baker and Wurgler (2002) repor a negaive relaion beween M/B efwa and curren leverage. They inerpre his relaion as consisen wih he hypohesis ha firms ime heir securiies issues and don subsequenly adjus heir leverage o he arge, so ha he iming impac persiss. Subsequen papers (e.g. 14

15 Kayhan and Timan (2007) Hovakimian (2006)) argue ha he observed negaive relaion is due o he informaion abou growh opporuniies conained in M/B efwa. In order o disinguish beween hese wo compeing inerpreaions, I run Baker and Wurgler s (2002) leverage regressions wih he weighed average of he hree componens of he M/B. If he negaive relaion beween M/B efwa and leverage arises because of cumulaive marke iming aemps hen one will observe ha Fmis efwa or Smis efwa are negaively relaed o leverage. If he negaive relaion comes hrough he variable Growh efwa hen one will conclude ha M/B efwa is a proxy for growh opporuniies and he persisence hypohesis will be rejeced Descripive saisics Table 2 repors he number of IPOs in he final IPO-sample, by year (Panel A.) and by indusry (Panel B.) over he period No surprisingly, he bulk of he IPO aciviy occurred in he lae nineies and in he echnology indusries. Nearly 50% of he IPO aciviy occurred beween 1995 and 2000 and nearly 30% of he IPO firms belong o he echnology indusry (Fama and French indusry code 6). Afer 2000, he number of IPOs falls by 77%. Table 3 provides he ime-series averages of he regression coefficiens for he valuaion model (Eq. (6)). The levels of R² indicae ha he cross-secional variaion of he marke value, wihin an indusry and for a given year, is highly explained by he accouning variables of he model. The average esimaed coefficiens display a variaion across he differen indusries. Uiliies have he lowes and he highes 1 j, consisenly wih hese firms 0 j having lower levels of inangibles. Conversely, he opposie characerisic is shared by he medicine and echnology indusries for insance. 15

16 The characerisics of he IPO firms a he pre-offering fiscal year-end and over he subsequen years are repored in Table 4 Panel A. Panel B. repors he characerisics for he calendar-year sample. As documened in previous sudies, firms experience a significan drop of heir leverage a he IPO year. Panel A. shows ha leverage sharply decreases from 64.4 percen a he end of he pre-ipo year o 35.5 a he end of he IPO year. I hen slighly increases over he subsequen years. The financing aciviy a he IPO year is noeworhy. Firms raised huge amouns of equiy: on average ne equiy issued represens 63.5 percen of heir asses. The equiy raised hen, conribued o he repaymen of deb, as he ne deb issuance corresponds o percen of asses. In comparison, Baker and Wurgler (2002) documen 32 percen of ne equiy issuance a he IPO year and 0.6 percen of ne deb issuance for heir sample firms. Firms rely more on deb finance in he year following he IPO bu deb finance declines in he subsequen years. The able shows ha equiy issuance is he main source of exernal finance, which is obviously in conradicion wih he pecking order hypohesis. Anoher feaure is he negaive sign of newly reained earnings. This may be aribued o firms from he echnology indusries ha accumulae losses because of he creaion of inangibles. Bu i may also be consisen wih he view ha periods of high levels of demand for IPOs also arac lower qualiy firms. Panel B shows ha newly reained earnings are posiive before In erms of valuaion variables, consisen wih previous sudies, firms experience a sharp decline of heir M/B raio a he IPO year. Bu his decline is mainly driven by he firmspecific misvaluaion componen ha drops from 1.55 o The secor-misvaluaion componen also declines in he year following he IPO. However, he growh componen rises a he following year and seadily declines hereafer. The calendar year panel shows ha he secor-misvaluaion componen perfecly reflecs ho and cold periods. I peaks during he 16

17 period Over he nex period, he M/B declines as well as is secor-misvaluaion componen. However, he firm-specific componen is he highes over ha period where ineresingly, besides issuing equiy firms pay back heir debs and build financial slack. These findings are clearly supporive of he marke iming heory. 4. Empirical findings 4.1. Marke valuaions around securiy issues In his secion I presen firms characerisics regarding heir marke-o-book raio and is hree componens around securiy issues. The purpose here is o gain insigh ino he abiliy of he error componens o serve as marke iming proxies. To do so, Table 5 compares he valuaions prevailing a he IPO year o he valuaions over he pos-ipo years (Column 1) and o he valuaions of a sample of non-ipo firms (Column2). The non-ipo sample comprises all firms of he calendar-year sample excep IPO-firms a heir IPO year. Table 5 shows ha he average M/B is he highes a he IPO year wih 2.70, whereas i drops o 1.98 on average over he subsequen years 5. In comparison, he average raio for non-ipo firms is The error componens and he long-run value o book componen have a similar behavior o ha of he M/B raio. They peak a he IPO year and are significanly lower aferward. These characerisics sugges ha he iniiaion of an IPO can be moivaed by firmspecific overvaluaion, by high indusry valuaions as well as high levels of growh opporuniies. Herzel and Li (2007) documen similar characerisics for SEO firms. RKRV acknowlege ha firm-specific error can be inerpreed eiher as misvaluaion or as firmspecific deviaions from conemporaneous indusry-average growh and discoun raes. 5 The means repored in his able differ slighly from hose repored in Table 4. This is because in his Table, firms are no required o have available daa on all he variables repored in Table 4. 17

18 Herzel and Li (2007) provide evidence on he effeciveness of he error componens o reflec misvaluaion raher han such deviaions. I ackle his issue in more deail, in secion In a recen sudy, Hovakimian (2006) ess wheher firms ime heir equiy and deb issues o equiy marke condiions by comparing he M/B raio a he beginning of he even year o he average M/B raio prevailing hree years prior and hree years afer he even year. His findings sugges ha firms end o issue equiy when heir M/B raios are unusually high bu fail o find a similar iming paern for deb issuers. I run a similar se of ess using he M/B raio and is componens. The objecive here is o examine if he paerns documened above for he IPO year hold for oher equiy issues and if deb issues exhibi iming paerns once marke condiions are measured by he misvaluaion componens of he M/B raio. A firm is defined as issuing equiy (deb) when he raio of ne equiy issues (deb) over oal asses (e/a) (d/a) exceeds 5%. Table 6 repors he means of he M/B raio and is componens from hree years prior o he even year, o hree years afer. The able shows a significan run-up of he M/B and each of he hree componens prior o an equiy issue. However, wih he excepion of he firm-misvaluaion componen, all he oher valuaion variables exhibi a significan higher level a he end of he issuance year han a he beginning of he even year. The secor misvaluaion componen is even higher in he pos-issuance period, suggesing ha firms issue equiy when he marke becomes ho, bu no a paricularly high levels of indusry valuaions. The firm misvaluaion componen as well as he long-run value o book componen experience a subsanial decline afer he equiy issuance. These findings sugges ha firms issue equiy during ho periods, when heir overvaluaion peaks and a high levels of growh opions. These resuls are consisen wih marke iming, and wih he abiliy of he firm-specific misvaluaion componen o properly capure he iming aemps. Besides, he resuls poin o an imporan shorcoming of using M/B raio as a 18

19 measure of marke iming as i is obvious ha he flucuaions of he M/B raio are highly driven by he levels of growh opporuniies. Regarding deb issues, he evidence on marke iming wih respec o M/B raio is mixed, and consisen wih he findings of Hovakimian (2006). While he mean pre-deb issue M/B is lower han he mean M/B raios prevailing in he previous years, i is also lower han is levels over he pos-even years. I is ineresing o noe ha he secor error componen evolves in a similar manner han around equiy issues. This rend implies ha when indusry valuaions increase, no only firms issue equiy bu also deb. This puzzling paern is unlikely o induce a negaive relaion whih leverage bu i may explain he resuls documened by Ali (2006) who finds ha firms ha go public in ho markes issue imporan amouns of deb afer he IPO and rebalance he impac of ho markes wihin a shor ime period. The GROWTH componen evolves in he same way han he M/B raio: a he ime of he issuance, firms experienced a decline in he value of heir growh opions, bu his decline persiss aferward. Ineresingly, he firm-specific misvaluaion componen drops in he deb issuance year and he rend significanly reverses righ afer he issuance year. Alhough he pos-issue decline is no monoonic, he level prevailing a he issue year remains he lowes over he period [-3; +3]. These findings offer furher evidence on he abiliy of he firmspecific misvaluaion componen o accuraely capure iming aemps and on he impac of perceived misvaluaion on leverage. Overall, he resuls ha emerge from his secion suppor he view ha he observed behavior of he M/B raio are likely o obain due o growh opporuniies which mask he effeciveness of he iming of he deb issues, for insance Shor erm impac of marke iming on capial srucure 19

20 The negaive impac of marke iming on annual changes in leverage has been widely documened in previous sudies. This secion analyses how he differen componens of M/B raio relae o changes in leverage and wheher heir impac, if any, comes hrough equiy issues, reained earnings or oher changes in asses. To separaely evaluae he effecs of growh opions and misvaluaion, I run he following regression ha conrols for oher deerminans of leverage and indusry fixed effes, in IPO-ime. D / A D / A FMIS SMIS GROWTH PPE/ A EBITDA/ A SIZE 7 2 D/ A (16) 3 Tangible asses may be associaed wih posiive changes in leverage as hey may be used as collaeral. The peching order heory predics ha firms wih high levels of inernal funds, as proxied by profiabiliy, end o have lower leverage raios. Size may be associaed wih lower leverage if large firms face less informaion asymery and herfore can issue more equiy. Large firms may also be less likely o ener financial disress leading size o be associaed wih higher levels of deb financing. Lagged leverage is included o conrol for mecanical backward moves of leverage when i reaches one of is boudaries. Panel A of Table 7 repors he regression coefficiens for Eq. (16) and for he regression using M/B insead of is componens for comparison purposes. The ne impac of growh opions on annual changes in leverage is significanly negaive. Even afer conrolling for his variable, he ne effec of firm-specific misvaluaion is o lower leverage, suggesing ha he well documened negaive relaionship beween M/B and leverage comes hrough growh opporuniies bu also hrough perceived specific overvaluaion. Ineresingly, he effec of secor misvaluaion is negaive and highly significan a he IPO year, bu i reverses in he following year and becomes insignifican aferward. This evidence complemens he

21 resuls documened by Ali (2006) for ho IPO-marke firms. The able also shows ha angibiliy induces increase in leverage. Profiabiliy ends o lower leverage, and he effec of size is o increase leverage bu his effec is significan only in he IPO year. The subsequen panels in Table 7 repor he regression coefficiens for he componens of he change in leverage over he same se of explanaory variables. The annual change in leverage can be decomposed as follows: D A D / A e / A RE / A E (1/ A 1/ A ). (17) / 1 Eq The firs erm in Eq. (17) is he negaive of ne equiy issues. If firms ime he equiy marke hen he negaive relaion observed beween misvaluaion and leverage should come hrough equiy issues. The second erm is he negaive of newly reained earnings. I allows o es wheher he decline in leverage is due o increased reained earnings. The las erm is he residual change in leverage which depends on oal growh in asses from he combinaison of equiy issues, deb issues, and newly reained earnings. Panel B. shows ha he hree componens of he M/B raio are posiively and significanly relaed o equiy issues. The resuls repored in Panel C. show ha firm-specific misvaluaion is posiively relaed o newly reained earnings bu only in few years and his effec is small in comparison o he increase in ne equiy issues. The negaive relaion beween he secor error componen of M/B and newly reained earnings is noeworhy. As discussed above, his relaion may be explained by he relaive high number of IPOs of firms from he echnology indusry druing he period of his analysis. This negaive relaion is responsible for he posiive or insignifican relaion beween secor overvaluaion and changes in leverage. Finally, Panel D. shows ha he M/B raio and is hree componens are generally posiively relaed o he residual changes in leverage, bu his is likely o induce an increase in leverage. The resuls regarding he oher variables are consisen wih previous sudies. 21

22 Overall, he resuls indicae ha he impac of firm-specific misvaluaion on changes in leverage is srong, and comes hrough equiy issues. The resuls also show an imporan impac of growh opporuniies on changes in leverage, consisen wih he predicions of radiional heories of capial srucure. However, he radeoff heory predics ha a high M/B raio is a sign of high levels of growh opions and ha he negaive effec of M/B on leverage is due o growh firms endency o use more equiy financing. Therefore, according o his heory, conrolling for growh opporuniies should resul in an insignifican relaion beween he remaining componens of M/B and ne equiy issues. The evidence presened in his secion is in clear opposiion wih he predicions of he radeoff heory, as he wo componens of misvaluaion are srongly posiively relaed o ne equiy issues Persisence of marke iming on capial srucure The evidence documened so far, is consisen wih a srong shor-erm impac of marke iming on capial srucure. Wheher his impac is persisen and impacs leverage on he long-run is he main quesion of he debae. Baker and Wurgler (2002) show ha he impac of marke iming, measured by M/B efwa, lass abou en years. Subsequen sudies find ha his effec is no persisen and ha he negaive relaion beween M/B efwa and leverage is likely o obain due o he informaion abou growh opporuniies conained in Baker and Wurgler s (2002) measure of iming. For insance, Ali (2006) shows ha he impac of ho marke issues vanishes wihin wo years and Hovakimian (2006) finds ha M/B efwa is posiively relaed o various measures of invesmen and hereby argue ha i is a proxy for firms invesmen prospecs. In his secion, using he exernal finance weighed averages of he hree componens of M/B raio defined in Eq. (10), Eq. (11) and Eq. (12), I invesigae he 22

23 relaionship beween make iming and capial srucure in a manner ha is unlikely o be affeced by growh opporuniies. Finally, as a robusness check, I quesion he abiliy of my iming variables o properly capure iming aemps Leverage regressions Table 8 repors resuls from he following OLS regression of leverage in IPO-ime and for he sample: D / A GROWTH 6 FMIS SMIS 2 PPE/ A EBITDA/ A SIZE Eq. (18) 8 GROWTH 3 9 FMIS 4 SMIS 5 Peersen (2005) suggess ha in he presence of a firm effec, he Rogers (1993) sandard errors clusered by firm are more accurae ha he OLS sandard errors. Therefore I adjus he -saisics for clusering by firm in he pooled sample (Column 5). The lagged values of he misvaluaion variables and he growh variable are included o conrol for heir conemporaneous correlaions wih leverage. The following paerns emerge from Table 8. Firs, he exernal weighed average firmspecific misvaluaion componen is negaive and highly significan over he en-year period following he IPO and over he enire period. Alhough he lagged FMIS is negaive and significan a some years, is magniude is lower han he magniude of FMIS efwa. This resul suggess ha he impac of marke iming on capial srucure remains highly persisen, even afer specifically conrolling for he effec of growh opporuniies. I is herefore obvious ha Baker and Wurgler (2002) resuls regarding he M/B efwa are o some 23

24 exen driven by marke iming and no only by growh opporuniies. However, in conras wih Baker and Wurgler (2002) resuls, he hisorical misvaluaion variable is no always he mos economically imporan variable. Growh opporuniies, wheher measured by he lagged or he hisorical variable exhibi a srong negaive effec on leverage. The impac of he imeseries secor error remains puzzling, consisenly wih he discussions above. The effec of he exernal weighed average secor-error is eiher posiive and significan or insignifican. To gain furher insigh on he pos-ipo financing policy of ho-indusries firms, I regressed ne deb issues 6 on he same se of variables I used for he change in leverage regressions. The resuls show ha firms wih a high secor-error componen of M/B issue subsanially more deb in he years following he IPO han oher firms. This paern, along wih he above documened negaive newly reained earnings, offses he negaive impac of equiy issues on leverage for hese firms. The iniial impac on leverage a he IPO year is even reversed aferward. This resul is again consisen wih he findings of Ali (2006) and poin o he inabiliy of hese variables o adequaely capure misvaluaion. Recall ha he firm-specific misvaluaion componen I use o measure he perceived misvaluaion capures he deviaions of he firm s marke value from is value implied by conemporaneous secor muliples. Regarding he secor error, RKRV ackowledge ha he calculaion of his variable requires he use of ex pos knowledge abou valuaion levels o infer misvaluaion. Therefore, he inerpreaion of his variable as a proxy for misvaluaion does no res on he inabiliy of marke paricipans o make full use of available informaion. Misvaluaion conained in his variable can no be aribued o behavioral biases bu o asymmeric informaion beween managers and invesors. On he oher hand, he calculaion of he firm-specific error 6 For brievey, he resuls are no repored bu are available upon reques. 24

25 componen uses informaion, unkown o he exernal invesors while available a he ime of he calculaion. I herefore capures perceived mispricing in he sense of he marke iming heory. Table 8 also indicaes ha he coefficiens displayed by he oher variables are generally consisen wih findings in previous sudies. Tangibiliy is posiively relaed o leverage, consisen wih he view ha angible asses may poenially serve as collaeral and hereby faciliae deb financing. Profiable firms have relaively lower levels of leverage as hey have, and larger firms have higher levels of leverage. Unrepored analysis using M/B efwa ha aemp o replicae he resuls of Baker and Wurgler (2002) provide similar resuls as hose documened by he auhors. However, I find a negaive and significan coefficien for M/B, ha may be due o he differences beween our wo samples. To gain furher insigh ino he persisen impac of misvaluaion on capial srucure, following Baker and Wurgler (2002), I run regressions of he cumulaive changes in leverage from he pre-ipo value on he same se of conrol variables as above plus he pre-ipo value of leverage. D / A D / A GROWTH 6 pre ipo 1 FMIS 7 0 SIZE SMIS 2 EBITDA/ A PPE/ A D/ A Eq. (19) 1 9 GROWTH 3 10 FMIS 4 pre ipo 1 SMIS If marke iming has a permanen effec on capial srucure, hen is iniial impac should be refleced in he cumulaive changes in leverage from is pre-ipo level, in he pos-ipo years. Table 9 shows ha he impac of marke iming lass for a leas en years afer he IPO. Even if here is a sligh decrease afer he year IPO+1, he magniude of he coefficien of 5 1 FMIS efwa is subsequenly sronger and increases from year o year as iming aemps accumulae. The hisorical secor-error variable shows a negaive and significan impac unil 25

26 IPO+1 and hen is overall impac on cumulaive changes in leverage becomes insignifican. The impac is even posiive in IPO+5. This resul is consisen wih he oher resuls discussed above regarding his variable. The iming of equiy issues due o high levels of growh opporuniies also shows a persisence. Overall, he resuls documened in his secion provide evidence on a long lasing impac of marke iming aemps on capial srucure. This evidence is consisen wih recen findings of Huang and Rier (2007) who demonsrae ha he effecs of pas securiies issues are long-lasing because firms adjus very slowly oward arge leverage Invesmen and cash regressions In his secion, I run robusness ess o assess he validiy of he misvaluaion and growh variables I used in he previous secion. Following Hovakimian (2006) I analyse wheher he negaive relaion beween my iming measure is a proxy for a firm s se of invesmen opporuniies, by examining is impac on curren invesmen. Table 10 repors he resuls from he following regressions: INV 7 0 CASH FMIS 1 8 SIZE SMIS 9 2 GROWTH 3 FMIS EBITDA A PPE/ A D/ A Eq. (20) / SMIS 5 GROWTH 6 INV is invesmen and is measured by capial expendiures o oal asses in Panel A., as he raio of R&D expenses o oal asses in Panel B. and as he change in non-cash asses o oal asses. I use he same regressors as Hovakimian: size, angibiliy, profiabiliy plus financial slack: CASH and lagged leverage. I include FMIS efwa, SMIS efwa and GROWTH efwa insead of 26

27 M/B efwa and FMIS, SMIS, GROWTH insead of M/B. Following Hovakimian, I include slack o conrol for he possibiliy ha he iming measure may be correlaed o curren invesmen because successful pas marke imers may have accumulaed cash ha can be used o fund curren invesmen opporuniies. I also include lagged leverage o conrol for he possibiliy ha marke imers may have lower curren levels of leverage and hence can inves more by raising deb. The regressions also include Fama and French 11 indusry dummies. Eq. (20) is esimaed in IPO-ime and over he sample period For he IPO-ime regressions - saisics use heeroscedasic consisen sandard errors. T-saisics are adjused for clusering in he pooled regressions. The hree Panels of able 10 show ha here is no relaion beween FMIS efwa and he differen invesmen measures. This iming variable is negaively relaed o he growh in noncash asses. I is neiher significanly relaed o capial expendiures nor o R&D expenses, excep for he year IPO+7 where i has a posiive coefficien, bu wih a relaively weak magniude. The hisorical growh variable doesn appear o be correlaed wih he growh in non cash asses. However, i is generally posiively relaed o capial expendiures. In addiion, his variables has a srong posiive impac on invesmen when measured by R&D expenses. The resuls regarding he hisorical secor-error variable sugges ha his variable is unlikely o proxy for growh opions, excep for he pooled regression of capial expendiures. In unrepored analysis, I replicae he invesmen ess of Hovakimian by using M/B efwa. Accordingly, I find a posiive relaion beween M/B efwa and invesmen when measured by capial expendiures or R&D, as documened by Hovakimian. However i is imporan o noe ha when I include he indusry dummies, he relaion becomes insignifican, suggesing ha eiher Hovakimian (2006) obains his resuls because he underesimaed he imporance of indusry fixed effecs, or because of he differences beween our samples. 27

28 I finally examine he exen o which hese variable may be relaed o he level of cash. Pas successful marke imers are likely o have accumulaed larger amouns of cash relaively o oher firms. In conras, firms wih high levels of growh opporuniies ha issued equiy o fund heir financing needs are unlikely o have high levels of cash. I run he following regression o invesigae he impac of FMIS efwa, SMIS efwa, and GROWTH efwa on curren levels of cash and shor erm invesmens scaled by asses: CASH / A SMIS 5 FMIS GROWTH SMIS 7 2 SIZE 8 GROWTH 3 EBITDA/ A PPE/ A D/ A Eq. (21) 9 FMIS 4 10 Table 11 shows ha he impac of FMIS efwa is o significanly increase he amoun of cash as a porion of oal asses, whereas firms wih high hisorical levels of growh opporuniies have relaively lower levels of cash in he early period following he IPO. The coefficiens become insignifican afer he year IPO+5. The resuls documened in his secion regarding he invesmen and he cash regressions sugges ha he exernal weighed average firm-specific error componen of he M/B raio is an accurae measure of marke iming ha is no conaminaed by he firm s se of invesmen opporuniies. 5. Conclusion The marke iming heory developped by Baker and Wurgler (2002) suggess ha capial srucure is he cumulaive oucome of pas aemps o ime he marke. Baker and 28

29 Wurgler (2002) repor ha he impac of he exernal finance weighed average marke-obook raio on capial srucure is highly persisen. The objecive of his sudy is o invesigae wheher he observed relaionship resuls from he persisen impac of marke iming or because of growh opporuniies, as prediced by radiional capial srucure heories. Using a mehodology developed by Rhodes-Kropf, Robinson and Viswanahan (2005), I decompose he marke o book raio ino misvaluaion and growh componens o compue a iming variable ha is no conamined by growh opions. This variable is he exernal finance weighed average of he firm-specific mivaluaion componen of he marke-o-book raio. Using a more accurae proxy for misvaluaion, I documen a long lasing impac of marke iming on capial srucure. Furhermore, I specifically rule ou he possibiliy ha he evidence provided in his analysis is obained due o growh opporuniies as he iming measure I use is unrelaed o various measures of invesmen bu posiively relaed o he building up of financial slack. 29

30 REFERENCES Ali, A. "How Persisen Is he Impac of Marke Timing on Capial Srucure?" Journal of Finance, 61 ( 2006), Baker, M., and J. Wurgler. "Marke Timing and Capial Srucure." Journal of Finance, 57( 2002), Fama, E. F., and K. R. French. Tesing Tradeoff and Pecking Order Predicions Abou Dividends and Deb. Review of Financial Sudies, 15 (2002), Fama, E. F., and J. Macbeh. Risk reurn and equilibrium: empirical ess. Journal of Poliical Economy, 71 (1973), Felham, G.A., and J.A. Ohlson. Valuaion and clean surplus accouning for operaing and financial aciviies. Conemporary Accouning Research, 11 (1995), Flannery, M. J., and K. P. Rangan. Parial Adjusmen and Targe Capial Srucures. Journal of Financial Economics, 79 (2006), Graham, J. R., and C. Harvey. "The heory and pracice of corporae finance: evidence from he field." Journal of Financial Economics, 60 (2001), Herzel, M.G., and Z. Li. "Behavioral and Raional Explanaions of Sock Price Performance Around SEOS: Evidence from a Decomposiion of Marke-o-Book Raios". Arizona Sae Universiy Working Paper, (2007). Huang, R., Rier, J.R. Tesing Theories of Capial Srucure and Esimaing he Speed of Adjusmen. Universiy of Florida Working Paper, Hovakimian, A. Are Observed Capial Srucures Deermined By Equiy Marke Timing? Journal of Financial and Quaniaive Analysis, 41 (2006), Hovakimian, A.; T. Opler; and S. Timan. The Deb-Equiy Choice. Journal of Financial and Quaniaive Analysis, 36 (2001), Jung, K., Y-C. Kim, and R.M. Sulz. Timing Invesmen Opporuniies, managerial discreion, and he securiy issue decision, Journal of Financial Economics, 42 (1996), Jensen, M., and W. Meckling. Theory of he Firm: Managerial Behavior, Agency Coss, and Ownership Srucure, Journal of Financial Economics, 3 (1976), Kayhan, A., and S. Timan. Firms Hisories and Their Capial Srucures. Journal of Financial Economics, 8 (2007),

31 Leary, M. T., and M. R. Robers. Do Firms Rebalance Their Capial Srucure? Journal of Finance, 60 (2005), Loughran, T., and J.R. Rier. The New Issues Puzzle, Journal of Finance, 50 (1995), Lucas, D., and R. MacDonald. Equiy Issues and Sock Price Dynamics, Journal of Finance, 45 (1990), Marsh, P. The choice beween equiy and deb: An empirical sudy, Journal of Finance, 37 (1982), Myers, S.C. Deerminans of Corporae Borrowing, Journal of Financial Economics, 5, (1977), Myers, S. C. The capial srucure puzzle. Journal of Finance, 39 (1984), Myers, S. C., and N. S. Majluf. Corporae financing and invesmen decisions when firms have informaion ha invesors do no have. Journal of Financial Economics, 13 (1984), Pagano, M., F. Panea, and L. Zingales. Why do companies go public? An empirical Analysis, Journal of Finance, 53 (1998), Peersen, M. A. Esimaing Sandard Errors in Finance Panel Daa Ses: Comparing Approaches. Unpublished Norhwesern Universiy and NBER working paper, (2006). Rhodes-Kropf, M., D.T. Robinson, and S. Viswanahan. Valuaion waves and merger aciviy: he empirical evidence. Journal of Financial Economics, 77 (2005), Rier, J., and R. Warr. The decline of inflaion and he bull marke of Journal of Financial and Quaniaive Analysis, 37 (2002), Rogers, W. Regression sandard errors in clusered samplers. Saa Technical Bullein, 13 (1993), Shyam-Sunder, L., and S. Myers. Tesing Saic Tradeoff Agains Pecking Order Models of Capial Srucure. Journal of Financial Economics, 51 (1999), Sulz, R. Managerial discreion and opimal financing policies. Journal of Financial Economics, 26 (1990), Welch, I. Capial srucure and sock reurns, Journal of Poliical Economy 112 (2004), Whie, H. A Heeroscedasiciy-Consisen Covariance Marix Esimaor and a Direc Tes of Heeroscedasiciy. Economerica, 48 (1980),