Management Qualty, Fnancal and Investment Polces, and Asymmetrc Informaton Thomas J. Chemmanur * Imants Paegls ** and Karen Smonyan *** Current verson: December 2007 * Professor of Fnance, Carroll School of Management, Boston College, 440 Fulton Hall, Chestnut Hll, MA 02467. E-mal: chemmanu@bc.edu. Phone: (617) 552-3980. Fax: (617) 552-0431. ** Assocate Professor of Fnance, John Molson School of Busness, Concorda Unversty, 1455 de Masonneuve Boulevard West, Montreal, Quebec H3G 1M8, Canada. E-mal: paegls@jmsb.concorda.ca. Phone: (514) 848-2424, ext. 2904. Fax: (514) 848-4500. *** Assstant Professor of Fnance, Sawyer Busness School, Suffolk Unversty, 8 Ashburton Place, Boston, MA 02108. E-mal: ksmonya@suffolk.edu. Phone: (617) 973-5385. Fax: (617) 305-1755. For helpful comments or dscussons, we thank Mark Lu, Debarsh Nandy, Evgena Portnaguna, Antonette Schoar, Susan Shu, and Hassan Tehranan, semnar partcpants at Boston College, as well as conference partcpants at the 2005 Amercan Fnance Assocaton Meetngs, the 2007 European Fnance Assocaton Meetngs, and the 2004 Fnancal Management Assocaton Meetngs. Specal thanks to an anonymous JFQA referee, and the JFQA Managng Edtor Hendrk Bessembnder for varous helpful comments that sgnfcantly mproved the paper. We also thank Sandra Caraballo for research assstance. Thomas Chemmanur acknowledges fnancal support from a Boston College Research Incentve Grant.
Management Qualty, Fnancal and Investment Polces, and Asymmetrc Informaton Abstract We develop measures of the management qualty of frms and make use of a unque sample of hand-collected data to examne the relatonshp between the reputaton and qualty of a frm s management and ts fnancal and nvestment polces, a relatonshp that has so far receved lttle attenton n the lterature. We hypothesze that better and more reputable managers are able to convey the ntrnsc value of ther frm more credbly to outsders, thus reducng the nformaton asymmetry facng ther frm n the equty market. Gven ths, frms wth better and more reputable managements wll have more access to the equty market, so that we expect lower leverage ratos for these frms. In addton, they wll have less need to sgnal usng dvdends, so that they wll have lower dvdend payout ratos. Further, snce better managers are lkely to select better projects (havng a larger NPV for any gven scale) and mplement them more ably, hgher management qualty wll also be assocated wth hgher levels of nvestment. We present evdence consstent wth the above hypotheses. Our drect tests of the relatonshp between management qualty and asymmetrc nformaton also ndcate that hgher management qualty leads to a reducton n the extent of nformaton asymmetry facng a frm n the equty market. Keywords: Management Qualty; Captal Structure; Dvdend Polcy; Investment Polcy; Asymmetrc Informaton. JEL classfcatons: G24; G31; G32; G35.
Management Qualty, Fnancal and Investment Polces, and Asymmetrc Informaton I. Introducton The determnants of the fnancal polces of a frm have been the subject of consderable debate n recent tmes (see Welch (2004) for a recent example). Surprsngly, almost ffty years after the semnal papers by Modglan and Mller on the captal structure and the dvdend polcy of a frm (see Modglan and Mller (1958, 1961)) and the large academc lterature n corporate fnance subsequent to ther work, there s lttle agreement n the fnance lterature regardng the determnants of a frm s fnancal polces. Further, t s often acknowledged that there may be several mssng dmensons determnng a frm s fnancal and nvestment polces that have not been analyzed n the lterature (see, e.g., Gompers, Ish, and Metrck (2004)). One mssng factor that may be mportant n determnng a frm s fnancal and nvestment polces s the qualty and reputaton of ts management team. For example, the qualty of a frm s management has been wdely regarded by practtoners (lke venture captalsts and fnancal analysts) as an mportant measure of frm qualty and a predctor of ts future performance. Better and more reputable management may be able to certfy the qualty of ther frm to the fnancal markets, thus reducng the extent of nformaton asymmetry between frm nsders and outsders. Ths means that frms wth hgher qualty managements may be able to access the equty market more easly, whch may n turn affect ther fnancal polces. Further, better managements may also be more lkely to select better (hgher NPV) projects and mplement them more ably, thus affectng ther frm s nvestment polcy. However, relatvely lttle attenton has been pad n the academc lterature so far to the relatonshp between a frm s management qualty and ts fnancal and nvestment polces, perhaps due to dffcultes n 1
operatonalzng the noton of management qualty and reputaton and due to data lmtatons. 1 Our objectve n ths paper s to remedy ths gap n the lterature. We develop varous measures of management qualty and reputaton (see secton IV for a dscusson) and use a unque sample of hand-collected data on the management qualty to study the relatonshp between the management qualty of a frm and ts fnancal and nvestment polces. Gven that a reducton n the extent of asymmetrc nformaton a frm faces n the fnancal markets s an mportant channel through whch management qualty may nfluence ts fnancal and nvestment polces, we also drectly study the relatonshp between management qualty and asymmetrc nformaton. The sample of frms n whch we study the relatonshp between management qualty and fnancal and nvestment polces are those makng seasoned equty offerngs (SEOs), snce ths allows us to collect detaled nformaton regardng varous varables affectng a frm s management qualty from SEO prospectuses. 2 The ablty of frm management wth hgher qualty and reputaton to certfy frm value more effectvely and thus reduce the extent of asymmetrc nformaton facng the frm n the equty market may arse n the followng manner. Frst, senor members of a frm s management team buld up reputatonal captal over ther career, snce these senor managers are nvolved n repeated dealngs wth the fnancal markets as part of ther job (e.g., rasng bank fnancng or rasng captal through a publc equty or debt ssue). Second, these managers know that there s a sgnfcant probablty that they wll leave ther frm to jon another, and that ther future employer wll be nfluenced by ther reputaton n dealng wth the fnancal markets n decdng whether or not to hre them, as well as n decdng ther compensaton package. Thrd, managers know that overprcng or hypng ther frm s stock (or decevng the fnancal markets n other ways) may tarnsh ther personal reputaton n the equty market, thereby dmnshng ther future value n the labor market; the greater the personal reputaton a manager has at stake, the greater s the future loss from msprcng hs frm s equty. As a consequence of the above, better and more reputable managers are more lkely to prce ther frm s equty farly, so that such frms are faced 2
wth less nformaton asymmetry about frm value between nsders and outsders. 3 Reduced nformaton asymmetry, n turn, may nfluence varous aspects of a frm s fnancal and nvestment polces. In partcular, frms facng lower nformaton asymmetry may be more lkely to ssue equty (snce they are more lkely to get a far prce for ther stock). Ths mples that frms wth better and more reputable managements can be expected to have lower levels of leverage. Also, a lower extent of asymmetrc nformaton facng the frm makes sgnalng to the equty market through dvdend polcy less mportant. Therefore, frms wth better and more reputable managements are lkely to have lower dvdend payout ratos. Better and more reputable managers may also be able to select better projects, characterzed by a larger NPV for any gven scale. Ths means that hgher management qualty frms are lkely to have a larger equlbrum scale of nvestment, whch, n turn, wll be reflected n hgher levels of captal expendtures. 4 We dscuss n more detal how management qualty and reputaton may affect a frm s fnancal and nvestment polces n the next secton. In order to ascertan whether reducng the asymmetrc nformaton facng a frm s ndeed an mportant channel through whch management qualty affects ts fnancal polces, we also drectly test whether management qualty s nversely related to the extent of asymmetrc nformaton facng the frm. To accomplsh ths, we use varous proxes for asymmetrc nformaton, lke the number of analysts followng the frm, analysts forecast error, dsperson of analysts forecasts, and the average bd-ask spread of a frm s stock. Further, snce lower the nformaton asymmetry facng a frm n the equty market, the greater the lqudty of the frm s equty, we also expect management qualty to be postvely related to lqudty, as measured by proxes such as tradng volume and share turnover. 5 The results of our emprcal tests ndcate that the qualty and reputaton of a frm s management ndeed affect ts fnancal and nvestment polces. Frst, we fnd that management qualty and reputaton are negatvely related to a frm s leverage and dvdend payout ratos. Second, management qualty and reputaton are postvely related to nvestment levels of the frm. 3
We fnd ths relatonshp between the management qualty and the fnancal and nvestment polces of a frm to be robust, snce ths relatonshp perssts even after controllng for potental smultanety and endogenety problems. Fnally, we fnd drect evdence that frms wth better and more reputable managements do ndeed face a lower extent of nformaton asymmetry n the equty market, and are characterzed by greater lqudty of ther shares, thus ndcatng that one of the channels through whch the management qualty of a frm affects ts fnancal and nvestment polces s by reducng the asymmetrc nformaton t faces n the equty market. In our formal emprcal analyss, we have focused on only two channels through whch the qualty and reputaton of a frm s management team can affect ts fnancal and nvestment polces. The frst channel s the reducton n the asymmetrc nformaton facng the frm n the equty market due to havng a hgh qualty management team. The second channel s the enhanced qualty of projects chosen by frms wth more able management teams, who may also mplement these projects more effcently. However, we cannot rule out the possblty of management qualty affectng a frm s fnancal and nvestment polces through other channels such as a reducton n the agency and bankruptcy costs facng the frm or through ncreasng the value of the frm s debt tax sheld: see secton VIII for a more detaled dscusson. Further, our emprcal desgn suffers from two lmtatons. Frst, due to data avalablty consderatons our sample conssts of only frms that have made a seasoned equty ssue (see footnote 2) whch makes t a rather unque sample. Second, our multvarate emprcal analyss may be subject to an endogenety problem, n the sense that n addton to management qualty affectng a frm s fnancal and nvestment polces, frms wth certan fnancal and nvestment polces may attract hgh qualty managers. Whle we have made every attempt to control for the above endogenety problem usng nstrumental varables n our regresson analyss, t can be argued that our choce of nstruments may not be approprate. We dscuss these two lmtatons of our emprcal desgn n secton IX. Gven the above lmtatons of our analyss, t s not our vew that we are able to dentfy causal relatonshps, or that we are able to analyze all possble channels through whch 4
the qualty and reputaton of a frm s management team may affect ts fnancal and nvestment polces. However, at a very mnmum, our emprcal study presents some very nterestng correlatons between the qualty and reputaton of a frm s management team and ts fnancal and nvestment polces, whch wll potentally nspre future researchers to nvestgate the underlyng relatonshps at a greater depth. 6 Our paper contrbutes to the newly emergng lterature on the relatonshp between the nature of a frm s management and varous aspects of ts fnancal and nvestment polces and performance. Chevaler and Ellson (1999) study the relatonshp between the performance of mutual funds and the characterstcs (age, experence, educaton, and SAT test scores) of ther fund managers. They fnd that managers who attended hgher-sat undergraduate nsttutons had sgnfcantly hgher rsk-adjusted excess returns. Bertrand and Schoar (2003) fnd that manager fxed effects explan some of the heterogenety n nvestment, fnancal, and organzatonal practces of frms. They attrbute these fxed effects to the dfferences n style across managers and lnk them to observable manageral characterstcs, such as the age of the managers and the nsttuton from whch they obtaned an MBA degree. 7 However, by focusng on specfc managers, they capture the effects of the management styles of varous managers on frms fnancal polcy, rather than the relatonshp between management qualty per se and fnancal polcy, whch s the topc we study here. Chemmanur and Paegls (2005) study the relatonshp between a frm s management qualty and varous aspects of ts IPO, and show that management qualty affects the characterstcs of a frm s IPO. 8 They, however, do not study the relatonshp between the management qualty of a frm and ts fnancal and nvestment polces, whch s the focus of ths paper. In summary, ths s the frst paper n the lterature to study the relatonshp between the qualty and reputaton of a frm s management and ts fnancal and nvestment polces. Ths s also the frst paper to drectly test the relatonshp between the qualty of a frm s management and the extent of asymmetrc nformaton facng the frm n the equty market, and the relatonshp between management qualty and the lqudty of ts stock. 5
The rest of the paper s organzed as follows. Secton II dscusses the relevant theory regardng the relatonshp between the management qualty and reputaton of a frm and ts fnancal and nvestment polces. Secton III summarzes the hypotheses we test n ths paper based on the theoretcal arguments made n secton II. Secton IV develops varous measures of management and frm qualty. Secton V descrbes our data and sample selecton procedures. Secton VI presents our emprcal testng methodology and reports our emprcal results regardng the relatonshp between management qualty and fnancal and nvestment polces. Secton VII drectly tests the relatonshp between the management qualty of a frm and the extent of asymmetrc nformaton facng t n the equty market. Secton VIII dscusses channels other than a reducton n asymmetrc nformaton through whch a frm s management qualty may affect ts fnancal polces and relates them to our emprcal fndngs. Secton IX dscusses some lmtatons of our emprcal desgn. Secton X concludes wth a summary of our results and our nterpretaton of them. II. How Does Management Qualty Affect a Frm s Fnancal and Investment Polces? We now dscuss n more detal the relatonshp between the qualty and reputaton of a frm s management and ts fnancal and nvestment polces. Earler, we argued that frms wth better management qualty and reputaton are lkely to face a reduced extent of nformaton asymmetry n the equty market. In order to concretze how managers wth hgher qualty and reputaton can help a frm reduce the nformaton asymmetry t faces n the equty market, consder the followng ntutve model. The model has three dates (tme 0, 1, and 2) and s analogous to the nvestment bank reputaton model of Chemmanur and Fulgher (1994) (see Fgure 1). At tme 0 (begnnng of perod one) the CEO (manager) of a frm has nosy nformaton superor to equty market nvestors about ts future cash flows, and makes an announcement (whch may or may not be truthful) to the equty market about ths cash flows (Step 1). The manager then chooses a fnancal polcy varable (e.g., leverage or dvdends) and announces ths 6
to the equty market as well, after whch equty market nvestors value the frm (Step 2). The manager s salary n the frst perod wll be an ncreasng functon of the frm s stock prce at ths tme. Subsequent to ths, the true value of the frm s future cash flows (announced earler by the manager) s realzed and revealed to equty market nvestors, who use ths to update the manager s reputaton for honesty usng Bayes Rule (Step 3). At the end of perod one (tme 1) the manager leaves the frm and jons a new frm, and Steps 1, 2, and 3 are repeated for the new frm n perod two. The game ends at tme 2 (.e., at the end of perod two). [Insert Fgure 1 here] In the above settng, the manager, though self-nterested, can act as an ntermedary to reduce nformaton asymmetry between the frm and equty market nvestors (analogous to an nvestment bank n Chemmanur and Fulgher (1994)), due to hs concern for hs reputaton for honesty. On the one hand, the manager has the ncentve to announce the hghest possble value for hs frm s future cash flows, snce ths can potentally ncrease the frm s stock prce at the end of perod one and hs personal compensaton n that perod. On the other hand, dong so would ncrease the probablty of hs cash flow estmate beng wrong and the chance of damagng hs personal reputaton for honesty (and consequently hs compensaton n hs future employment). 9 In the above settng, t can be shown that, the greater the current reputaton of the manager, the more lkely he s to tell the truth about the frm s future cash flows. Snce equty market nvestors are aware of ths, the more reputable the manager, the more credble they wll fnd hm, and consequently the more effectve he wll be n mtgatng the nformaton asymmetry between the frm and the equty market (for a roadmap to formally demonstratng the above results, see Chemmanur and Fulgher (1994)). In summary, the hgher the qualty and reputaton of a frm s management team at the tme the frm announces the choce of a fnancal polcy varable, the lower the extent of nformaton asymmetry the frm wll face n the fnancal market. We now dscuss how a reducton n the nformaton asymmetry facng a frm due to better management qualty and reputaton may affect ts fnancng and dvdend polces. Frst, 7
consder the relatonshp between nformaton asymmetry and leverage. As shown by Myers and Majluf (1984), equty s the fnancng vehcle of last resort when a frm faces nformaton asymmetry n the captal markets (see also Gammarno and Lews (1989), Noe (1989), and Chemmanur and Jao (2005)). However, n the presence of other market mperfectons (e.g., costs of fnancal dstress assocated wth debt), a frm may rase a sgnfcant porton of fnancng by ssung equty n equlbrum snce t trades off the fnancng costs arsng from nformaton asymmetry wth those arsng from these other mperfectons. In such a settng, the lower the extent of asymmetrc nformaton facng a frm, the greater the equlbrum fracton of external fnancng rased through equty. Thus, ths theory predcts that, ceters parbus, frms wth better management qualty and reputaton wll be assocated wth lower equlbrum leverage ratos. Now consder the relatonshp between management qualty and dvdend polcy. A number of papers n the theoretcal corporate fnance lterature have argued that the motvaton for a frm to pay out dvdends s to sgnal frm nsders prvate nformaton to the equty market (see, e.g., Bhattacharya (1979), Mller and Rock (1985), and John and Wllams (1985)). In the above settng, the greater the extent of nformaton asymmetry facng the frm n the equty market, the stronger the sgnal that needs to be sent to dsspate ths nformaton asymmetry, so that the amount of dvdends pad by a frm s larger n equlbrum. In summary, ths theory predcts that, ceters parbus, frms wth better management qualty and reputaton wll be assocated wth lower equlbrum dvdend payout ratos. We now dscuss the effect of management qualty on the nvestment polcy of a frm. Management qualty may affect a frm s nvestment polcy drectly through manageral ablty. If better and more reputable managers are able to select better projects for ther frms (characterzed by a larger NPV for any gven scale), and assumng decreasng returns to scale, a frm wth better management qualty wll be assocated wth a larger equlbrum scale of nvestment (see Fgure 2). A larger scale of nvestment, wll, n turn, be reflected n hgher levels of captal expendtures, so that these wll also be ncreasng n a frm s management qualty. [Insert Fgure 2 here] 8
III. Summary of Hypotheses As we argued above, the qualty and reputaton of a frm s management may affect ts fnancal and nvestment polces n a varety of ways. We now summarze the hypotheses arsng from the above arguments, whch wll serve as the bass for our emprcal tests. Our frst set of hypotheses deals wth the relatonshp between the qualty and reputaton of a frm s management and ts fnancal and nvestment polces: H1: The leverage rato of a frm wll be negatvely related to qualty and reputaton of ts management. 10 H2: The dvdend payout rato of a frm wll be negatvely related to qualty and reputaton of ts management. H3: Frms wth better and more reputable managements wll have hgher levels of nvestment (captal expendtures). Our second set of hypotheses deals wth the drect relatonshp between management qualty and the extent of asymmetrc nformaton facng the frm, and between management qualty and lqudty: H4: The degree of asymmetrc nformaton facng a frm wll be negatvely related to ts management qualty and reputaton. H5: The lqudty of a frm s equty wll be postvely related to ts management qualty and reputaton. IV. Measures of Management Qualty and Reputaton, and Frm Qualty A. Measures of Management Qualty and Reputaton In ths secton we descrbe our measures of management qualty and reputaton whch we measure along the followng three broad dmensons: management team resources, management team structure, and management reputaton n the busness communty. To measure each of these dmensons, we make use of several dfferent varables. 9
Management team resources refer to the human and knowledge resources (ncludng both educaton and relevant work experence) avalable to frm management. Ths s measured by the number of executve offcers (defned as those wth a rank of vce presdent or hgher) on a frm s management team (TSIZE). Team resources also depend upon knowledge and educaton of ts members, whch we measure n two ways. Our second measure of team resources s the percentage of the management team members wth an MBA degree (PMBA). We also measure the percentage of the management team members who are Certfed Publc Accountants (PCPA). Hgher percentages of MBAs and CPAs mply hgher management qualty. The thrd factor that ncreases management team resources s a pror top management experence. We look at the percentage of the management team members who have served as executve offcers and/or vce presdents or hgher pror to jonng the frm (PFTEAM). The greater the value of ths varable, the better the management qualty. Fnally, management team resources are also enhanced by the presence of managers who are n charge of the frm s core functonal areas. Our ffth measure, therefore, s the percentage of the management team members n the core functonal areas: operatons and producton, research and development, marketng and sales, and fnance (CORE). Clearly, hgher values of CORE mean better management qualty. The second dmenson along whch we measure management qualty, namely, management team structure, reflects the extent of unformty or heterogenety n the tenures of management team members, and ther relatve mportance to the management team. We look at the average tenure of the management team members (TENURE) defned as the average number of years team members have been wth the team. Hgher average tenure may ndcate coheson and shared experences, and thus would mply lower transacton costs among team members. However, long tenures mght also ndcate the presence of complacency and rgdfyng effects on team nteractons. An deal team would have members from dfferent cohorts, whch would allow for an nflow of new deas and perspectves. Thus, a hgher management qualty 10
would be assocated not only wth a long average tenure, but also wth a hgher dsperson of tenures. Our second measure of team structure, therefore, s tenure heterogenety, defned as the coeffcent of varaton of the team members tenures (TENHET). The thrd measure of team structure we use s CEO domnance. On the one hand, a strong-wlled and domnatng chef executve offcer may severely dmnsh possble contrbutons from subordnate team members. On the other hand, a strong CEO mght mprove the coheson of the management team. Thus, whle we beleve that CEO domnance s an mportant measure of team qualty, we are agnostc about the drecton of the expected mpact (postve or negatve) of ths measure on management qualty. Our measure of CEO domnance s the rato of CEO salary and bonus n the fscal year precedng SEO to the average salary and bonus of other team members lsted n the executve compensaton secton of the prospectus (FCEO). Assumng that CEOs have a substantal nfluence over ther own pay and nearly total nfluence over ther subordnates pay, ths measure reflects the gap between the CEO s assessment of hs own worth to the frm and hs assessment of other team members worth, and s thus a good measure of CEO domnance. 11 The thrd dmenson along whch we measure management qualty, namely, management reputaton n the busness communty, reflects the reputaton bult up by members of the management team. Ths s measured by the number of corporate boards (apart from that of ther own frm) that management team members st on (BOARDS). Whle the measures dscussed above also partally measure management reputaton, ths varable s a better ndex of the reputaton and vsblty of a frm s management n the busness communty. Agan, the greater the value of ths varable, the hgher the qualty and reputaton of a frm s management. Table 1 summarzes our measures of management qualty and reputaton, as well as dependent varables that wll be used n subsequent regressons. The average (medan) management team sze (TSIZE) was 6.45 (6) wth the smallest management team consstng of only one person, and the largest one of 18 members. On average 6.8 percent of managers had an 11
MBA degree (PMBA), 7.5 percent were CPAs (PCPA), 44.1 percent had held a top management poston at another frm pror to jonng the SEO frm (PFTEAM), and 54.7 percent held postons n the core functonal areas (CORE). The average tenure (TENURE) ranged from 1 to 26.7 years, wth a mean (medan) of 5.02 (4.29) years. 12 On average, CEOs were earnng 47.6 percent more than the average member of ther management team (FCEO). [Insert Table 1 here] Though the above varables are expected to measure management qualty and reputaton, each of them may have unque lmtatons as a measure of the underlyng unobservable construct. Therefore, we use common factor analyss to construct a sngle varable for each management qualty dmenson descrbed above that captures varaton common to the observable proxes of management qualty. 13 In order to ensure that varous factors are capturng only the effects of management qualty (and not those of frm sze or frm age) we use frm-sze- and frm-ageadjusted proxy varables to extract the factors. Team resources factor (TRF) score s obtaned usng common factor analyss on the frm-sze-adjusted TSIZE, PMBA, PCPA, PFTEAM, and CORE varables. Team structure factor (TSF) score s obtaned usng common factor analyss on the frm-age-adjusted TENURE, TENHET, and frm-sze-adjusted FCEO varables. 14 Table 2 reports the results of the above common factor analyss. Panel A presents startng communaltes, calculated as the squared multple correlatons obtaned from regressng each of the management qualty measures on the other measures wthn the same dmenson, whle Panel B reports the egenvalues of the reduced correlaton matrces. As suggested by Harman (1976), the number of factors needed to approxmate the orgnal correlatons among ndvdual measures s equal to the number of summed egenvalues needed to exceed the sum of communaltes. In our sample the summed communaltes are less than the egenvalues for the frst factor n the factor analyss for each dmenson of management qualty and reputaton, suggestng that one factor n each of the dmensons parsmonously explans the ntercorrelatons among the ndvdual measures. Correlatons between the common factor scores and ther respectve orgnal measures 12
of management qualty are reported n Panel C, whle Panel D reports summary statstcs of the team resources factor (TRF) and the team structure factor (TSF) scores. [Insert Table 2 here] B. Proxes for Other Aspects of Frm Qualty In order to separate the effects of management qualty and reputaton from those of other dmensons of frm qualty, we control for these other dmensons usng the followng proxes. Frst, we use two frm qualty varables commonly used n the fnance lterature: frm sze and frm age. We use the natural log of the book value of assets (LNBVA) as a proxy for the frm sze. 15 Frm age s defned as the natural log of one plus the number of years the frm has return data avalable from CRSP (LFAGE). The larger and older the frm, the hgher the frm qualty. 16 Second, we control for the number of outsde drectors, defned as the number of drectors lsted n the management secton of the prospectus who are not executve offcers of the frm (ODIR). There are two ways n whch the number of outsde drectors affects frm qualty. Frst, outsde drectors may provde addtonal knowledge (nputs and perspectves) to the frms managers. Second, they provde lnks to external partes, such as audtors, fnancal nsttutons, and underwrters. The greater the number of outsde drectors, the better the frm qualty. 17 V. Data and Sample Selecton The data used n ths study came from several dfferent databases. The lst of seasoned equty offerngs between 1993 and 1997 was obtaned from SDC/Platnum Global New Issues Database. After elmnatng non-underwrtten ssues, ssues wth offer prces of less than $5, spnoffs, fnancal frms (all frms wth SIC codes between 6000 and 6999), and foregn frms we were left wth 2,006 frms. Further, 32 frms were elmnated because they dd not have the necessary nformaton on CRSP/Compustat. Of the remanng companes, 6 dd not have prospectuses and 338 dd not have management nformaton avalable n ther prospectuses. After elmnatng these frms, we were left wth 1,630 frms n our sample. 13
The data on management qualty was hand-collected from the prospectuses of frms conductng SEOs. The prospectuses were obtaned from Thomson Fnancal Database. In partcular, nformaton on the team sze, tenure, former work experence, level of educaton, and board membershps of the team members came from the management secton of the prospectuses. The data necessary to calculate the CEO domnance (FCEO) were obtaned from the executve compensaton sectons of the prospectuses. Informaton on stock prces was obtaned from CRSP, whle the accountng nformaton was obtaned from Compustat. Fnally, the data necessary to calculate asymmetrc nformaton varables was obtaned from IBES, whle the data on bd-ask spreads was obtaned from CRSP and supplemented by TAQ database. VI. Management Qualty, and Fnancal and Investment Polces In ths secton we dscuss the emprcal methodology used to test our hypotheses and report our results of the relatonshp between management qualty and fnancal and nvestment polces. Frst we present the results of our unvarate and multvarate tests regardng ths relatonshp. Then we present the results of our robustness tests of the same relatonshp. A. Unvarate Tests Table 3 reports the results of our unvarate tests of the relatonshp between management qualty and fnancal and nvestment polces of a frm. We splt the sample nto two by the medan management qualty factor scores (TRF and TSF). We wll refer to the frms wth below medan management qualty factor scores as low management qualty frms, and those wth above medan management qualty factor scores as hgh management qualty frms. [Insert Table 3 here] We measure a frm s fnancal and nvestment polces usng the followng varables. Frst, we measure a frm s leverage (LEVER) as the rato of long-term debt (Compustat tem 9) plus debt n current labltes (Compustat tem 34) to the sum of the long-term debt, the debt n 14
current labltes, and the book value of common equty (Compustat tem 60). Second, we measure a frm s dvdend payout rato (DIVID) as the rato of the sum of common and preferred dvdends (Compustat tems 21 and 19, respectvely) to the earnngs before deprecaton, nterest, and taxes (Compustat tem 13). Thrd, we measure a frm s nvestment level (INVEST) as the rato of captal expendtures (Compustat tem 128) to the total assets at the begnnng of the year (Compustat tem 6). All the precedng varables are for the fscal year of the SEO. We fnd that the team resources factor score (TRF) has a sgnfcant nfluence on fnancal and nvestment polcy varables. In partcular, we fnd that hgh management qualty frms (.e. frms wth an above medan TRF score) have lower leverage and dvdend payout ratos, and hgher levels of nvestment. For example, frms wth hgh management qualty have, on average, leverage ratos that are lower than those of low management qualty frms by 3.8 percentage ponts (ths s a 16 percent decrease over the leverage rato of low management qualty frms). We also fnd statstcally sgnfcant dfferences n the medan levels of nvestment and dvdend payout ratos between hgh and low management qualty frms. The mpact of the team structure factor score (TSF) on fnancal polcy varables s ambguous. Whle we observe lower levels of leverage for frms wth better and more reputable managements (frms wth above medan TSF scores have, on average, leverage ratos that are lower than those of below medan TSF score frms by 2.3 percentage ponts), the results on dvdend payout ratos go the opposte way. B. Management Qualty and Leverage To test the hypothess that frms wth better and more reputable managements have lower leverage (hypothess H1), we run the followng regresson: (1) LEVER 6 = β TSIZE 12 0 PCPA ODIR 1 7 CORE 13 TOBINQ 2 TENURE 8 BOARDS 14 ROA 3 TENHET 9 15 FCEO RETSD 4 10 PMBA LNBVA + IND + ε, 5 PFTEAM 11 LFAGE + + 15
where TOBINQ s the Tobn s Q, defned as the market value of assets dvded by the book value of assets (Compustat tem 6), where market value of assets equals the book value of assets plus the market value of common equty (Compustat tem 25 tmes Compustat tem 199) less the sum of the book value of common equty (Compustat tem 60) and balance sheet deferred taxes (Compustat tem 74); ROA s the return on assets defned as the rato of earnngs before extraordnary tems (Compustat tem 18) over the total assets at the begnnng of the year (Compustat tem 6); RETSD s the standard devaton of the total stock return calculated over 255 tradng days for the fscal year of the ssue, and IND s a set of two-dgt SIC code dummes (used to account for varatons n fnancal polcy varables across ndustres). 18 The coeffcents on the frst eght ndependent varables are expected to be negatve, whle we are agnostc about the sgn of the coeffcent of FCEO. If a strong CEO dmnshes contrbutons from other team members, the coeffcent s expected to be postve, whle f the team coheson s ncreased due to a strong CEO, the coeffcent s expected to be negatve. The remanng ndependent varables are unrelated to management qualty and are used to control for other aspects of frm qualty. 19 Regressons 1 to 4 n Table 4 present the results of the regressons wth leverage as a dependent varable. We fnd that frms wth hgher percentages of MBAs (PMBA) and managers n the core functonal areas (CORE) have sgnfcantly lower leverage. In partcular, a one standard devaton ncrease n CORE leads to a 1.5 percentage pont decrease n the leverage rato (about 14.3 percent of the medan frm s leverage rato), whle a one standard devaton ncrease n PMBA results n a 0.7 percentage pont decrease n leverage rato (about 6.8 percent of the medan frm s leverage). In regresson 4 we replace the ndvdual management qualty varables wth team resources factor (TRF) and team structure factor (TSF) scores. We fnd that TRF score has a statstcally sgnfcant negatve mpact on leverage wth ts coeffcent sgnfcant at the 1 percent level. Whle TSF score also affects a frm s leverage negatvely, ts coeffcent s not statstcally sgnfcant. [Insert Table 4 here] 16
C. Management Qualty and Dvdend Payout Ratos To test the effect of management qualty and reputaton on dvdend payout ratos (hypothess H2), we estmate the followng regresson: (2) DIVID 6 12 = β 0 PCPA ODIR TSIZE 1 7 CORE 13 TOBINQ 2 TENURE 8 BOARDS 14 ROA 3 TENHET 9 + IND FCEO + ε. 4 10 PMBA LNBVA 5 PFTEAM 11 LFAGE + + We expect management qualty and reputaton to have a negatve effect on the dvdend payout ratos. Therefore we expect negatve coeffcents on the frst eght measures of management qualty and are agnostc about the sgn of the coeffcent of FCEO. As reported n regressons 5 to 8 n Table 4, the percentages of management team wth a pror top management experence (PFTEAM) and n the core functonal areas (CORE) have a negatve effect on dvdend payout ratos. In regresson 8 we use TRF and TSF scores nstead of ndvdual management qualty varables. The TRF score has a statstcally sgnfcant negatve mpact on a frm s dvdend payout rato wth ts coeffcent sgnfcant at the 5 percent level. The coeffcent of the TSF score s also negatve though not sgnfcant. The results are also economcally sgnfcant. A one standard devaton ncrease n PFTEAM results n a 0.26 percentage pont (about 23.1 percent of the average frm s payout rato) decrease n the dvdend payout rato and a one standard devaton ncrease n CORE results n a 0.17 percentage pont (about 15.0 percent of the average frm s payout rato) decrease n the dvdend payout rato. Our fndngs of a negatve relatonshp between management qualty and dvdend payout ratos are consstent wth the noton that better qualty managers reduce nformaton asymmetry between the frm and the equty market, whch makes sgnalng through dvdends less relevant. 20 D. Management Qualty and the Level of Investment To test the relatonshp between management qualty and reputaton and the level of a frm s nvestment (hypothess H3), we run the followng regresson: 17
(3) INVEST 6 12 = β 0 PCPA ODIR TSIZE 1 7 13 CORE 2 TOBINQ TENURE 8 BOARDS 14 ROA 3 TENHET 9 + IND FCEO + ε. 4 10 PMBA LNBVA 5 PFTEAM 11 LFAGE + + We expect the coeffcents of the frst eght management qualty proxes to be postve, and we are agnostc about the sgn of the coeffcent of FCEO. Regressons 1 to 5 n Table 5 report the results of our regressons wth nvestment level as a dependent varable. We fnd that frms wth larger management teams (TSIZE), hgher percentages of CPAs (PCPA) on the management team, and hgher percentages of managers n the core functonal areas (CORE) have sgnfcantly hgher levels of nvestment. In regresson 5 we use TRF and TSF factor scores nstead of ndvdual management qualty varables. The team resources factor score has a statstcally sgnfcant postve effect on a frm s nvestment level wth ts coeffcent sgnfcant at the 1 percent level. [Insert Table 5 here] The nfluence of management qualty varables on the nvestment level s also economcally sgnfcant. A one standard devaton ncrease n the TSIZE leads to a 1.4 percentage pont (about 13.8 percent of a medan frm s nvestment level) ncrease n the level of nvestment. Also, one standard devaton ncreases n the percentage of CPAs (PCPA) and the percentage of managers n the core functonal areas (CORE) lead to 1.8 and 1.1 percentage pont ncreases, respectvely, n the nvestment level. Our results are consstent wth the noton that better and more reputable managers are better at selectng and mplementng new projects and, therefore, have hgher equlbrum levels of nvestment. E. Robustness Tests of the Relatonshp between Management Qualty and Fnancal Polces It can be argued that not only management qualty affects the fnancal and nvestment polces of frms but frms wth certan fnancal and nvestment polces attract hgh qualty managers; n other words, management qualty and fnancal polces are determned smultaneously. In ths case the OLS estmators from prevous sectons may suffer from 18
smultanety problems: ndependent varables of management qualty may be correlated wth the error terms of the above regressons. To address ths problem, we make use of the seemngly unrelated regresson (SURE) framework where we estmate each fnancal and nvestment polcy regresson equaton (equatons (1) through (3) n prevous sectons) smultaneously wth two other equatons, where management qualty factor scores serve as dependent varables whle the respectve fnancal and nvestment polcy varables serve as ndependent varables together wth other frm qualty varables. We rely on the covarance matrx estmated from OLS results and then estmate the three regresson equatons smultaneously. For example, for the relatonshp between leverage and management qualty, we estmate a system of three smultaneous equatons: (4) LEVER = β0 1BOARDS 2LNBVA 3LFAGE 4ODIR 5TOBINQ 6ROA 7RETSD 8TRF 9TSF + IND + ε, TRF = β0 1LNBVA 2LFAGE 3ODIR 4TOBINQ 5ROA + 6LEVER + IND + ε, TSF = β0 1LNBVA 2LFAGE 3ODIR 4TOBINQ 5ROA + 6LEVER + IND + ε. + The relatonshp between management qualty and other fnancal polcy and nvestment varables s estmated n a smlar manner. The results of ths estmaton are presented n Table 6. For each fnancal and nvestment polcy varable we smultaneously estmate three regressons, where n regresson 1 the fnancal or nvestment polcy varable s the dependent varable and n regressons 2 and 3 management qualty factor scores TRF and TSF are the dependent varables, respectvely, as explaned above. The coeffcent of the team resources factor (TRF) score has the predcted sgn and s hghly sgnfcant n all the regressons wth LEVER, DIVID, and INVEST as dependent varables. Ths provdes further support to our fndngs n the unvarate tests and OLS regressons that management qualty s negatvely related to the leverage and dvdend payout ratos whle postvely related to the level of nvestment. The effect of management qualty measured by the team structure factor (TSF) score on fnancal and nvestment polces s less pronounced as documented prevously n the unvarate tests and OLS regressons: the TSF 19
coeffcent has the predcted sgn n LEVER and DIVID regressons (though not sgnfcant), and ts sgn s opposte to that predcted n the INVEST regresson. [Insert Table 6 here] Further, t can be argued that management qualty s endogenous n the equatons of the relatonshp between fnancal and nvestment polces and management qualty (equatons (1) through (3) n prevous sectons). To address ths endogenety ssue we make use of the two-stage OLS regresson methodology, where n the frst stage we regress management qualty factor scores on the explanatory varables and use the predcted values of management qualty factor scores from the frst stage as ndependent varables n the second stage regresson where fnancal and nvestment polcy varables are the dependent varables. The choce of nstrumental varables s dffcult n ths context, snce we need to keep n mnd that the set of nstrumental varables should nclude at least some varables that are correlated wth the management qualty factors and vrtually uncorrelated wth fnancal and nvestment polcy varables. We use underwrter reputaton (REP) and resdual standard devaton (RESSD) as nstrumental varables. REP s the natural logarthm of one plus the lead underwrter s reputaton as measured by Loughran and Rtter (2004); t has relatvely low correlaton wth all fnancal polcy varables (the hghest correlaton coeffcent s that wth leverage and equal to 0.09) and has greater correlaton wth management qualty (the correlaton coeffcent wth the TRF score s 0.18). RESSD s the resdual standard devaton estmated usng market model over 255 tradng days for the fscal year of the ssue. The results of our two-stage OLS regressons are presented n Table 7. The coeffcents of the team resources factor (TRF) score have the predcted sgn and are sgnfcant n the regressons wth LEVER and DIVID as dependent varables. The coeffcent of the TRF score has the predcted sgn n the INVEST regresson as well, though not sgnfcant. The coeffcents of team structure factor (TSF) score are nsgnfcant n all the regressons though wth the predcted sgn n LEVER and INVEST regressons. The results of our two-stage OLS estmaton are broadly consstent wth those of the SURE estmaton. 21 [Insert Table 7 here] 20
VII. Management Qualty, Asymmetrc Informaton, and Lqudty In ths secton we present the results of our drect tests regardng the relatonshp between management qualty and the level of asymmetrc nformaton about the frm. We use two groups of asymmetrc nformaton measures. The frst group ncludes the number of analysts followng the frm, the analysts forecast error, the dsperson of forecasts, and the bd-ask spread. 22 The second group ncludes two lqudty measures: tradng volume and equty turnover. 23 A. Unvarate Tests Table 8 presents our results of unvarate tests of the relatonshp between management qualty and the degree of asymmetrc nformaton about the frm n the equty market. We splt the sample nto two by the medan management qualty factor scores (TRF and TSF). [Insert Table 8 here] We measure the degree of asymmetrc nformaton facng the frm n the equty market usng the followng varables. Frst, the number of analysts (NUMEST) s the natural logarthm of one plus the number of analysts followng the frm. Second, the forecast error (FORERR) s the rato of the absolute dfference between the mean analysts EPS estmate and the actual EPS to the stock prce. Thrd, the dsperson of forecasts (DISPERS) s the rato of the standard devaton of analysts forecasts to the stock prce. All of the values are for the end of the fscal year pror to the SEO. Fourth, bd-ask spread (BIDASK) s the medan monthly rato of bd-ask spread to the stock prce (calculated over the fscal year pror to the ssue). Ffth, tradng volume (PVOL) s the natural logarthm of the product of the stock prce and the monthly tradng volume, averaged over the fscal year pror to the offer. Sxth, share turnover (TO) s the rato of the monthly tradng volume to the number of shares outstandng, averaged over the fscal year precedng the offer. We fnd that management qualty and reputaton do ndeed reduce the degree of asymmetrc nformaton between the frm and the equty market. In partcular, frms wth an above medan team resources factor score (TRF) have a greater analyst coverage, lower forecast 21
error, lower bd-ask spread, and are more lqud. For example, hgh management qualty frms have 0.78 more analysts followng them, a 0.4 percentage pont reducton n analysts forecast error, and a 0.5 percentage pont reducton n the bd-ask spread relatve to the correspondng values for low management qualty frms. Ths corresponds to a 25 percent ncrease n the number of analysts, a 40 percent decrease n analysts forecast error, and a 12 percent decrease n the bd-ask spread for hgh management qualty frms relatve to the correspondng values for low management qualty frms. Also, average tradng volume of hgh management qualty frms s $5.5 mllon (a 68 percent ncrease) hgher, whle ther share turnover s approxmately 3.7 percentage ponts hgher (a 29 percent ncrease) than those of low management qualty frms. All these dfferences are statstcally sgnfcant at the 1 percent level, except for the forecast error whch s sgnfcant at the 5 percent level. We fnd that the team structure factor (TSF) score has a smlar, though somewhat less sgnfcant nfluence on the degree of asymmetrc nformaton. Frms wth above medan TSF scores have more analysts followng them and have lower dsperson and bd-ask spread. The dfferences between hgh and low management qualty frms are also economcally sgnfcant. For example, frms wth better qualty managements have 0.23 more analysts followng them, whch s around 10 percent more than the number of analysts followng low management qualty frms. Also, the bd-ask spread of hgh management qualty frms s lower than that of low management qualty frms by 0.2 percentage pont (a 20 percent decrease). B. Management Qualty and Asymmetrc Informaton To test the hypothess that frms wth hgher management qualty and reputaton have lower degree of asymmetrc nformaton (hypothess H4), we run the followng regresson: (5) NUMEST = β 5 10 0 PFTEAM LNBVA TSIZE 1 6 11 PCPA 2 LFAGE TENURE 7 CORE 12 ODIR 3 TENHET 8 + ε. BOARDS 4 PMBA 9 FCEO + + 22
We expect the coeffcents of the frst eght management qualty varables to be postve, whle we are agnostc about the sgn of the coeffcent of FCEO. Regressons 1 to 5 n Panel A of Table 9 report the results of the regressons wth the number of analysts as the dependent varable. We fnd that TSIZE, TENURE, CORE, and BOARDS have a sgnfcantly postve nfluence on the number of analysts followng the frm. In regresson 5 we use TRF and TSF scores nstead of ndvdual management qualty varables: both scores have a sgnfcantly postve mpact on the number of analysts followng a frm. These relatonshps are also economcally sgnfcant. A one standard devaton ncrease n CORE ncreases the number of analysts followng the company by 0.31 (31% of the number of analysts that follow the medan frm). Also, one standard devaton ncreases n TSIZE and BOARDS lead to ncreases n the number of analysts by 0.23 and 0.13, respectvely. Next, n regresson equaton (5) we substtute number of analysts by FORERR. We expect management qualty to have a negatve nfluence on the forecast error. Agan, we are agnostc about the sgn of the coeffcent of FCEO. Regressons 6 to 8 n Panel A of Table 9 report the results of our regressons wth forecast error as the dependent varable. We fnd that frms wth longer average tenures (TENURE) and hgher percentages of CPAs (PCPA) have lower forecast errors. Coeffcents of these varables are sgnfcant at the 5 percent level. A one standard devaton ncrease n PCPA leads to a 0.17 percentage pont decrease n the forecast error (about a 22 (77) percent decrease n the average (medan) frm s forecast error). In regresson 8 we replace ndvdual management qualty varables by TRF and TSF factor scores. Though both scores have negatve coeffcents, they are not statstcally sgnfcant. [Insert Table 9 Panels A and B here] We now substtute DISPERS as the dependent varable n regresson equaton (5). 24 We expect management qualty proxes to have a negatve effect on the dsperson. Regressons 1 to 4 n Panel B of Table 9 present results wth dsperson as the dependent varable. We fnd that frms wth longer tenures of ts managers (TENURE) and hgher percentages of CPAs (PCPA) have 23
sgnfcantly lower dspersons. These results are sgnfcant at the 5 and 1 percent levels, respectvely, and are also economcally sgnfcant. One standard devaton ncreases n TENURE and PCPA decrease the dsperson by 0.08 and 0.07 percentage ponts (67 and 58 percent of the medan frm s dsperson), respectvely. In regresson 4 we use TRF and TSF scores as measures of management qualty. Whle TSF score has a negatve coeffcent sgnfcant at the 10 percent level, the TRF score has a postve coeffcent and sgnfcant at the 5 percent level. Fnally, we substtute BIDASK as the dependent varable n regresson equaton (5). We expect better management qualty frms to have lower bd-ask spreads and therefore negatve coeffcents of management qualty varables. Regressons 5 to 8 n Panel B of Table 9 report the results of these regressons. We fnd that frms wth larger management teams (TSIZE) and hgher percentages of CPAs (PCPA) have sgnfcantly lower bd-ask spreads. One standard devaton ncreases n TSIZE and PCPA, respectvely, reduce the bd-ask spread by 0.34 and 0.18 percentage ponts (equvalent to 11 and 6 percent declnes n the medan frm s bd-ask spread). In regresson 8 we use TRF and TSF scores as measures of management qualty. Both scores affect bd-ask spread negatvely, wth the TRF coeffcent sgnfcant at the 1 percent level. These results confrm our conjecture that better and more reputable managements are able to mtgate the degree of nformaton asymmetry between the frm and the equty market. Ths relatonshp between management qualty and the extent of nformaton asymmetry facng the frm s both statstcally and economcally sgnfcant. C. Management Qualty and Lqudty We now study whether frms wth better and more reputable management have more lqud equty (hypothess H5) and run the followng regresson: (6) PVOL = β TSIZE 5 0 10 1 PFTEAM LNBVA 6 11 2 TENURE PCPA LFAGE 7 CORE 12 3 ODIR TENHET 8 BOARDS 13 4 PMBA FCEO SHAROUT + ε, 9 + + 24
where SHAROUT s the number of shares outstandng (n mllons) at the end of the fscal year precedng the offer. We expect postve sgns of the coeffcents of the frst eght management qualty proxes, but are agnostc about the sgn of the coeffcent of FCEO. Regressons 1 to 7 n Table 10 present the results of our regressons wth PVOL as the dependent varable. We fnd that frms wth larger management teams (TSIZE), hgher percentages of CPAs (PCPA) and managers n the core functonal areas (CORE) are sgnfcantly more lqud. The coeffcents of these varables are sgnfcant at the 1 percent level. In regresson 7 we replace ndvdual management qualty varables by TRF and TSF scores. Both scores postvely affect tradng volume wth the coeffcent of TRF sgnfcant at the 1 percent level. These results are also economcally sgnfcant. A one standard devaton ncrease n TSIZE brngs about $1.9 mllon (about 18% of the average frm s monthly tradng volume) ncrease n PVOL. Also, one standard devaton ncreases n CORE and PCPA lead to around $1.6 and $0.9 mllon ncreases n PVOL, respectvely. [Insert Table 10 here] Our second measure of lqudty s the average monthly turnover. To test f management qualty and reputaton s postvely related to lqudty (as measured by the monthly turnover), we replace PVOL wth TO n equaton (6) and exclude SHAROUT as a control varable. Regressons 8 to 12 n Table 10 report the results of varous specfcatons of the above regresson equaton. Our fndngs ndcate that frms wth larger management teams (TSIZE) and hgher percentages of managers n core functonal areas (CORE) have sgnfcantly hgher turnovers n ther equty. The coeffcents of these varables are sgnfcant at the 1 percent level. In regresson 12 we use TRF and TSF factor scores nstead of ndvdual varables of management qualty: the TRF score has a postve coeffcent whch s sgnfcant at the 1 percent level. The effects of TSIZE and CORE on the monthly share turnover are also economcally sgnfcant. One standard devaton ncreases n CORE and TSIZE result n 1.3 and 1 percentage pont ncreases, respectvely, n turnover. These ncreases represent around 12.4 and 9.0 percent ncreases, respectvely, n the medan frm s turnover. 25
The above fndngs of a postve relatonshp between the qualty and reputaton of a frm s management and the lqudty of ts equty provde addtonal evdence that hgher management qualty frms face a lower extent of asymmetrc nformaton n the equty market. VIII. Alternatve Channels through whch Management Qualty May Affect a Frm s Fnancal and Investment Polces In addton to the two channels that we have examned above, there are several others through whch the qualty and reputaton of a frm s management team may affect ts fnancal and nvestment polces. We brefly dscuss some of these channels below and relate them to our emprcal fndngs, though we wll leave t to future research to sort out these alternatve explanatons of the effect of a frm s management qualty on ts fnancal and nvestment polces. A. Management Qualty, Agency Costs, and Fnancal Polces Management qualty can affect leverage through the dscplnng role of debt n a prncpal-agent type settng (see, e.g., Grossman and Hart (1982)). Consder a stuaton where manageral effort ncreases frm value, and hgher qualty managers have a lower cost of effort. Ths means that, n the absence of debt, the equlbrum level of effort exerted by lower qualty management wll be lower than that exerted by hgher qualty management. Debt can dscplne management n ths stuaton f managers suffer personal penaltes n the event of bankruptcy. Snce the beneft of ths dscplnng role of debt wll be greater for lower qualty managers (and the costs of havng debt n a frm s captal structure are unaffected by management qualty), ths theory would predct (lke the asymmetrc nformaton based theores that we have examned n detal) that the equlbrum level of leverage wll be decreasng n management qualty. Thus, our emprcal fndng that frms wth hgher management qualty and reputaton are assocated wth lower leverage ratos s consstent wth the predcton of theores based on agency costs as well. 26
We now examne the predctons of agency cost based theores of dvdends (see, e.g., Easterbrook (1984)) for the relatonshp between management qualty and dvdend polcy. Lke asymmetrc nformaton based theores, such theores also mply that hgher management qualty wll lead to lower dvdend payout ratos. Under the agency cost argument, a hgh dvdend payout rato reduces managers ablty to waste a frm s free cash flow by nvestng n negatve net present value projects (see also Jensen (1986)). Such a hgh dvdend payout rato forces frms to rase external fnancng for nvestment n new projects, and snce external fnancng s subject to montorng by nsttutonal nvestors and other ntermedares n the captal market, a frm s lkely to be able to undertake only postve net present value projects usng such fnancng. However, rasng external fnancng more frequently requres the frm to ncur addtonal drect and ndrect costs: a frm s equlbrum dvdend payout rato emerges from tradng off the above benefts and costs of payng out a hgher proporton of ts cash flow as dvdends. Gven that managers of hgher qualty and reputaton are less lkely to engage n dsspatng a frm s free cash flow (see Holmström (1999) for a theoretcal analyss demonstratng that concern for reputaton mtgates the moral hazard problem facng a manager), the above argument mples that frms wth hgher management qualty wll have lower equlbrum dvdend payout ratos. Thus, our emprcal fndng that frms wth hgher management qualty and reputaton are assocated wth lower dvdend payout ratos are consstent wth the predctons of agency cost based theores of dvdends as well. B. Management Qualty, Bankruptcy Costs, and Fnancal Polces The qualty of a frm s management may affect ts leverage polcy through ts effect on expected bankruptcy costs. Consder a stuaton where hgher qualty frm management shfts a frm s cash flow dstrbuton to the rght by managng t more effectvely. Ths mples that, for any gven level of debt, the expected value of the bankruptcy cost faced by frms wth hgher qualty management teams wll be lower (snce ther probablty of bankruptcy wll be lower). 27
Assumng that the benefts of havng a gven level of debt n ther frm s captal structure (arsng from the nterest tax sheld: see next sub-secton) s the same (or greater) for frms wth hgher qualty management teams, ths mples that the equlbrum amount of debt n a frm s captal structure wll be ncreasng n ts management qualty. However, our emprcal fndng that frms wth hgher management qualty have less leverage s nconsstent wth the above predcton. C. Management Qualty, Taxes, and Fnancal Polces The qualty of a frm s management may affect ts leverage through ts effect on the magntude of the nterest tax sheld. Smlar to the stuaton dscussed n the prevous sub-secton, consder the case where hgher qualty frm management shfts a frm s cash flow dstrbuton to the rght by managng t more effectvely. Ths ncreases the ncome of the frm subject to corporate taxes, so that the tax benefts of havng a hgher level of debt wll be greater for frms wth hgher qualty managements. Assumng that the cost assocated wth havng any gven level of debt s ether the same or lower for frms wth hgher qualty managers (as we argued would be n the prevous sub-secton), ths mples that the equlbrum level of debt n a frm s captal structure wll be ncreasng n ts management qualty. However, our emprcal fndng that frms wth hgher management qualty have less leverage s nconsstent wth the above predcton. IX. Lmtatons of Our Emprcal Desgn A. Data Lmtatons One lmtaton of our emprcal desgn s that our sample conssts of frms makng SEOs, snce ths allows us to collect detaled nformaton regardng management qualty from SEO prospectuses (frms dsclose detaled nformaton regardng ther management team n ther IPO as well as SEO prospectuses). We choose to study frms makng SEOs rather than those makng IPOs, gven that such frms, beng more mature, wll have more stable fnancal polces compared to frms gong publc (whch may be n the early stages of defnng ther fnancal 28
polces). However, t can be argued that such frms are not representatve of the unverse of frms, snce we are condtonng on a specfc fnancng event (namely an SEO), lmtng the ablty to generalze the conclusons about the relatonshp between the management qualty of a frm and ts fnancal and nvestment polces arsng from our emprcal analyss. Gven the above lmtaton of our sample, our emprcal study can be vewed only as a frst step n analyzng the relatonshp between a frm s management qualty and ts fnancal and nvestment polces. B. Correctng for Endogenety n Management Qualty In order to control for the potental endogenety n management qualty n our regresson analyss of the relatonshp between a frm s management qualty and ts fnancal and nvestment polces, we make use of a two-step OLS regresson methodology. We make use of underwrter reputaton and resdual standard devaton as nstrumental varables n ths regresson analyss. The choce of nstrumental varables s dffcult n ths context, snce the set of nstrumental varables should nclude at least some varables that are correlated wth our management qualty factors but uncorrelated wth our fnancal and nvestment polcy varables. One can argue that our nstrumental varables may not satsfy the above requrements, snce underwrter reputaton or resdual standard devaton may affect a frm s fnancal and nvestment polces through varables other than ts management qualty. Gven ths, our emprcal analyss can be vewed only as a frst step n analyzng the relatonshp between a frm s management qualty and ts fnancal and nvestment polces. X. Summary, Interpretaton of Results, and Concluson Usng a sample of frms makng seasoned equty offerngs, we examne, for the frst tme n the lterature, the relatonshp between the reputaton and qualty of a frm s management and ts fnancal and nvestment polces and also the relatonshp between the management qualty and the extent of nformaton asymmetry facng the frm n the equty market. Our man fndngs 29
are as follows. Frst, the qualty of a frm s management s negatvely related to ts leverage and dvdend payout ratos. Second, management qualty and reputaton are postvely related to the level of a frm s nvestments. Thrd, we fnd drect evdence that management qualty and reputaton are negatvely related to the extent of asymmetrc nformaton facng the frm n the equty market, and postvely related to the lqudty of the frm s equty. Overall, our fndngs suggest that management qualty and reputaton are ndeed mportant dmensons of frm qualty, and that measures of these varables provde addtonal nformaton about the ntrnsc value of a frm over and above that provded by frm qualty varables such as frm age and frm sze. Our fndngs are consstent wth better and more reputable frm management beng able to convey the value of ther frm more credbly to outsders, thus reducng the nformaton asymmetry facng the frm n the equty market. In partcular, our fndngs that hgher management qualty and reputaton are assocated wth lower levels of leverage and lower dvdend payout ratos are consstent wth better managements facng a lower degree of asymmetrc nformaton n the equty market. Ths s also drectly supported by our tests of the relatonshp between management qualty and asymmetrc nformaton, whch document an nverse relatonshp between the qualty and reputaton of a frm s management and varous proxes for the extent of nformaton asymmetry facng the frm n the equty market. Further, snce better qualty projects chosen by hgher qualty managers have larger NPV for any gven scale, the equlbrum scale (level of nvestment) of such projects wll also be larger, requrng larger captal expendtures. Thus, the postve relatonshp we document between the qualty and reputaton of a frm s management and ts level of captal expendtures provdes evdence consstent wth the noton that frms wth hgher qualty managers select better projects and are able to mplement them more successfully. 30
References Bertrand, M., and A. Schoar. Managng wth Style: The Effect of Managers on Frm Polces. Quarterly Journal of Economcs, 118 (2003), 1169 1208. Bhattacharya, S. Imperfect Informaton, Dvdend Polcy, and the Brd n the Hand Fallacy. Bell Journal of Economcs, 10 (1979), 259 270. Borokhovch, K. A.; R. Parrno; and T. Trapan. Outsde Drectors and CEO Selecton. Journal of Fnancal and Quanttatve Analyss, 31 (1996), 337 355. Chemmanur, T. J., and P. Fulgher. Investment Bank Reputaton, Informaton Producton and Fnancal Intermedaton. Journal of Fnance, 49 (1994), 57 79. Chemmanur, T. J., and Y. Jao. Seasoned Equty Issues wth Soft Informaton: Theory and Emprcal Evdence. Workng Paper, Boston College (2005). Chemmanur, T. J., and I. Paegls. Management Qualty, Certfcaton, and Intal Publc Offerngs. Journal of Fnancal Economcs, 76 (2005), 331 368. Chevaler, J., and G. Ellson. Are Some Mutual Fund Managers Better Than Others? Cross- Sectonal Patterns n Behavor and Performance. Journal of Fnance, 54 (1999), 875 899. Chrste, A. A. On Cross-Sectonal Analyss n Accountng Research. Journal of Accountng and Economcs, 9 (1987), 231 258. Clarke, J., and K. Shastr. On Informaton Asymmetry Metrcs. Workng Paper, Unversty of Pttsburgh (2001). Cotter, J. F.; A. Shvdasan; and M. Zenner. Do Independent Drectors Enhance Target Shareholder Wealth Durng Tender Offers? Journal of Fnancal Economcs, 43 (1997), 195 218. Crémer, J. Corporate Culture and Shared Knowledge. Industral and Corporate Change, 2 (1993), 351 386. D Aven, R. A. Top Manageral Prestge and Organzatonal Bankruptcy. Organzaton Scence, 1 (1990), 121 142. Easterbrook, F. H. Two Agency-Cost Explanatons of Dvdends. Amercan Economc Revew, 74 (1984), 650 659. Eckbo, B. E., and Ø. Norl. Lqudty Rsk, Leverage, and Long-Run IPO Returns. Journal of Corporate Fnance, 11 (2005), 1 35. Gaver, J. J., and K. M. Gaver. Addtonal Evdence on the Assocaton Between the Investment Opportunty Set and Corporate Fnancng, Dvdend, and Compensaton Polces. Journal of Accountng and Economcs, 16 (1993), 125 160. Gammarno, R. M., and T. Lews. A Theory of Negotated Equty Fnancng. Revew of Fnancal Studes, 1 (1989), 265 288. 31
Gompers, P. A.; J. L. Ish; and A. Metrck. Incentves vs. Control: An Analyss of U.S. Dual- Class Companes. Workng Paper, Unversty of Pennsylvana Wharton School (2004). Grossman, S. J., and O. D. Hart. Corporate Fnancal Structure and Manageral Incentves. In The Economcs of Informaton and Uncertanty, J. J. McCall, ed. Chcago, IL: Unversty of Chcago Press (1982). Guay, W. R. The Senstvty of CEO Wealth to Equty Rsk: An Analyss of the Magntude and Determnants. Journal of Fnancal Economcs, 53 (1999), 43 71. Hambrck, D. C., and R. A. D Aven. Top Team Deteroraton as Part of the Downward Spral of Large Corporate Bankruptces. Management Scence, 38 (1992), 1445 1466. Harman, H. H. Modern Factor Analyss, 3 rd ed. Chcago, IL: Unversty of Chcago Press (1976). Hermaln, B. E. Economcs and Corporate Culture. In The Internatonal Handbook of Organzatonal Culture and Clmate, S. Cartwrght et al., eds. Chchester, UK: John Wley & Sons, Ltd. (2001). Hodgson, G. M. Corporate Culture and the Nature of the Frm. In Transacton Cost Economcs and Beyond, J. Groenewegen, ed. Boston, MA: Kluwer Academc Press (1996). Holmström, B. Manageral Incentve Problems: A Dynamc Perspectve. Revew of Economc Studes, 66 (1999), 169 182. Jensen, M. C. Agency Costs of Free Cash Flow, Corporate Fnance, and Takeovers. Amercan Economc Revew, 76 (1986), 323 329. John, K., and J. Wllams. Dvdends, Dluton, and Taxes: A Sgnalng Equlbrum. Journal of Fnance, 40 (1985), 1053 1070. Kreps, D. M. Corporate Culture and Economc Theory. In Perspectves on Postve Poltcal Economy, J. E. Alt and K. A. Shepsle, eds. Cambrdge, UK: Cambrdge Unversty Press (1990). Kreps, D. M., and R. Wlson. Reputaton and Imperfect Informaton. Journal of Economc Theory, 27 (1982), 253 279. Krshnaswam, S., and V. Subramanam. Informaton Asymmetry, Valuaton, and the Corporate Spn-Off Decson. Journal of Fnancal Economcs, 53 (1999), 73 112. Kyle, A. S. Contnuous Auctons and Insder Tradng. Econometrca, 53 (1985), 1315 1335. Lazear, E. P. Corporate Culture and the Dffuson of Values. In Trends n Busness Organzaton, H. Sebert, ed. Tubngen, Germany: J.C.B. Mohr (1995). Loughran, T., and J. R. Rtter. Why Has IPO Underprcng Changed Over Tme? Fnancal Management, 33 (2004), 5 37. Mchaely, R., and W. H. Shaw. The Prcng of Intal Publc Offerngs: Tests of Adverse- Selecton and Sgnalng Theores. Revew of Fnancal Studes, 7 (1994), 279 319. 32
Mller, M. H., and K. Rock. Dvdend Polcy under Asymmetrc Informaton. Journal of Fnance, 40 (1985), 1031 1051. Modglan, F., and M. H. Mller. The Cost of Captal, Corporaton Fnance and the Theory of Investment. Amercan Economc Revew, 48 (1958), 261 297. Modglan, F., and M. H. Mller. Dvdend Polcy, Growth, and the Valuaton of Shares. Journal of Busness, 34 (1961), 411 433. Myers, S. C., and N. S. Majluf. Corporate Fnancng and Investment Decsons When Frms Have Informaton that Investors Do Not Have. Journal of Fnancal Economcs, 13 (1984), 187 221. Noe, T. H. Captal Structure and Sgnalng Game Equlbra. Revew of Fnancal Studes, 1 (1989), 331 355. Rtter, J. R. The Hot Issue Market of 1980. Journal of Busness, 57 (1984), 215 240. Welch, I. Captal Structure and Stock Returns. Journal of Poltcal Economy, 112 (2004), 106 131. 33
1 Two notable exceptons to ths are Bertrand and Schoar (2003) and Chemmanur and Paegls (2005), as we dscuss n more detal below. 2 Frms dsclose detaled nformaton about ther management team n ther IPO and SEO prospectuses. We choose to study frms makng SEOs, snce such frms, beng more mature, wll have more stable fnancal polces than frms gong publc (whch may be n the early stages of defnng ther fnancal polces). 3 Chemmanur and Fulgher (1994) develop a theoretcal analyss demonstratng that reputable, long-lved players assocated wth a frm can help reduce the extent of nformaton asymmetry facng t n the equty market, thereby ncreasng the offer prce for ts equty. Analogously, reputaton of a frm s management team can help reduce the extent of nformaton asymmetry t faces n the equty market. 4 The qualty of a frm s management can also potentally affect ts fnancal and nvestment polces by nfluencng ts corporate culture. However, whle an emergng lterature has attempted to provde economc content to corporate culture (see, e.g., Kreps (1990), Crémer (1993), Lazear (1995), Hodgson (1996), and Hermaln (2001)), ths lterature does not generate clear testable predctons on the relatonshp between management qualty and a frm s fnancal and nvestment polces. 5 See, e.g., Kyle (1985), who shows that lqudty s nversely related to the amount of prvate nformaton exstng about the frm n the equty market. 6 An anonymous referee provded very nsghtful comments on the ssues dscussed n ths paragraph, for whch we are partcularly thankful. 7 In partcular, they fnd that CEOs wth MBA degrees and younger generatons of CEOs, on average, are more aggressve: they have hgher levels of debt, captal expendtures, and lower dvdend payout ratos. 8 Chemmanur and Paegls (2005) fnd that frms wth better managements have lower IPO underprcng and underwrtng expenses, have more reputable underwrters, and larger nsttutonal nterest n ther IPO. 9 As s standard n the fnte horzon reputaton modelng lterature (Kreps and Wlson (1982), Chemmanur and Fulgher (1994)) one can operatonalze the noton of manageral reputaton by postulatng a small fracton of managers who are pathologcally honest (.e., who always tell the truth). Whle the objects of our analyss are not the above knds of managers but those who may choose to tell the truth or otherwse based on ther own self-nterest, equty market nvestors would measure the reputaton of a manager by ther probablty assessment that he s of the pathologcally honest type. Clearly, the more accurate the 34
manager s forecast about the frm s future cash flows, the hgher the updated value of hs reputaton for honesty. Note however that, snce the manager s prvate nformaton at tme 0 about the frm s future cash flows s nosy, equty market nvestors are aware that he may make honest mstakes, so that a wrong forecast would not drve hs reputaton to zero. 10 We expect ths relatonshp (as well as others) to hold after controllng for frm sze and other frm qualty varables. 11 Smlar measures have been used n the strategy and organzatonal behavor lterature to study the nfluence of management team qualty and reputaton on frm performance: see, e.g., D Aven (1990) and Hambrck and D Aven (1992), who use such measures to study the deteroraton of management team qualty around bankruptces. See also Chemmanur and Paegls (2005). 12 One could argue that fnancal and nvestment polces are not a functon of management qualty and reputaton but rather are nherted from prevous managements as long-standng polces. However, the summary statstcs of our sample ndcate that, whle the average (medan) management team tenure s 5.02 (4.29) years, the average (medan) number of years snce lstng on an exchange s 5.16 (2) years. Ths mples that, on average, current management teams of frms n our sample were formed around the tme these frms went publc. Gven that many of a frm s fnancal and nvestment polces are decded subsequent to gong publc, t seems to be the case that current managers (whose qualty and reputaton are captured by our management qualty measures) ndeed had a domnant role n formulatng these polces. 13 A number of papers n the emprcal corporate fnance and accountng lterature make use of factor analyss to solate the unobservable construct underlyng several proxy varables. See, e.g., Gaver and Gaver (1993) and Guay (1999), who make use of factor analyss to study the sze of a frm s nvestment opportunty set. 14 Snce the unadjusted TENURE and TENHET varables are most hghly correlated wth frm age, we adjusted these varables for age. On the other hand, snce the CEO domnance varable (FCEO) s most hghly correlated wth frm sze, we adjusted ths varable for sze. 15 We use book value of assets as a measure of frm sze to be consstent wth our analyss of frms fnancal and nvestment polces where such polcy varables are constructed usng accountng nformaton and some of them (such as captal expendtures) are normalzed by the book value of assets. 35
16 These two measures of frm qualty have been wdely used n the lterature (see, e.g., Rtter (1984), Mchaely and Shaw (1994)). 17 Several studes n the corporate control lterature have shown that outsde drectors enhance frm value (see, among others, Cotter, Shvdasan, and Zenner (1997) and Borokhovch, Parrno, and Trapan (1996)). 18 We nclude TOBINQ and ROA as addtonal control varables n regresson equatons (1) through (3) to control for potental effect of frms nvestment opportunty sets and performance on fnancal and nvestment polcy varables. We also nclude RETSD n the regresson equaton (1) to control for frms asset rsk whch may affect ther leverage ratos. 19 In order to account for possble nteractons between varous fnancal and nvestment polcy varables, we also estmated regresson equatons (1) through (3) ncludng LEVER, DIVID, and INVEST n those regressons as ndependent varables. For example, n the leverage regresson we ncluded DIVID and INVEST as ndependent varables, n the dvdends regresson we ncluded LEVER and INVEST as ndependent varables, and, fnally, n the nvestments regresson we ncluded LEVER and DIVID as ndependent varables. The results of all these estmatons were smlar to the ones reported here. 20 Ths s also consstent wth hgher qualty managers havng better projects for ther frms, characterzed by a larger NPV for any gven scale, so that they renvest a larger fracton of cash flows that ther frms generate nto these postve net present value projects nstead of payng them out to shareholders. 21 We also use the three-stage OLS estmaton technque to address the endogenety and smultanety ssues at the same tme. We use the same nstrumental varables as n the two-stage regressons; the results of our three-stage OLS estmaton were smlar to our two-stage estmaton results and are avalable upon request. 22 These measures are smlar to those used by Chrste (1987), Krshnaswam and Subramanam (1999), and Clarke and Shastr (2001). 23 Eckbo and Norl (2005) use these lqudty measures to study lqudty rsk and long-run IPO returns. 24 We also used the standard devaton of analysts forecasts as the dependent varable. The results are qualtatvely unchanged wth ths dependent varable. 36
Table 1 Summary statstcs. TSIZE s the sze of a frm s management team, defned as executve offcers wth a rank of vce presdent or hgher. PMBA s the percentage of a frm s management team wth MBA degrees. PCPA s the percentage of a frm s management team who are certfed publc accountants. PFTEAM s the percentage of the management team who have served as executve offcers and/or vce presdents pror to jonng the SEO frm. CORE s the percentage of the management team havng core functonal expertse, namely, holdng postons n operatons, sales and marketng, research and development, and fnance. TENURE s the average number of years management team members have been wth the team. TENHET s the coeffcent of varaton of the team members tenures. FCEO s the rato of CEO salary and bonus n the fscal year precedng the SEO to the average salary and bonus of other management team members. BOARDS s the number of other companes boards that management team members st on. LNBVA s the natural logarthm of the book value of the frm s assets at the end of the fscal year pror to the SEO. LFAGE s the frm age defned as the natural logarthm of one plus the number of years the frm has return data avalable from CRSP. ODIR s the number of outsde drectors. TOBINQ s the market value of assets dvded by the book value of assets, where market value of assets equals the book value of assets plus the market value of common equty less the sum of the book value of common equty and balance sheet deferred taxes. ROA s the rato of earnngs before extraordnary tems to the book value of total assets. RETSD s the standard devaton of the total stock return calculated over 255 tradng days for the fscal year of the ssue. LEVER s the leverage, defned as long-term debt plus debt n current labltes over long-term debt plus debt n current labltes plus book value of common equty. DIVID s the dvdend payout rato defned as the sum of common and preferred dvdends over earnngs before deprecaton, nterest, and taxes. INVEST s the nvestment level calculated as the rato of captal expendtures to book value of assets at the begnnng of the fscal year. NUMEST s the natural logarthm of one plus the number of analysts followng the frm at the end of the fscal year precedng the ssue. FORERR s the forecast error, defned as the rato of the absolute dfference between analysts mean EPS estmate and actual EPS to the stock prce. DISPERS s the rato of the standard devaton of analysts forecasts to the stock prce. BIDASK s the medan monthly bd-ask spread as a percentage of the stock prce, calculated over the fscal year precedng the ssue. PVOL s the natural logarthm of the product of the stock prce and the monthly tradng volume, averaged over the fscal year precedng the offer. TO s the rato of monthly tradng volume to the number of shares outstandng, averaged over the fscal year precedng the offer. LEVER, DIVID, and INVEST are wnsorzed at the 99 th percentle. For LEVER and DIVID only postve observatons are reported. Panel A: Summary statstcs of ndependent varables mn mean medan max std. dev TSIZE 1 6.453 6 18 2.556 PMBA 0 0.068 0 0.833 0.141 PCPA 0 0.075 0 1 0123 PFTEAM 0 0.441 0.429 1 0.285 CORE 0 0.547 0.571 0.900 0.203 TENURE 1 5.017 4.286 26.667 3.128 TENHET 0 0.684 0.671 1.741 0.300 FCEO 0 1.476 1.446 4.438 0.447 BOARDS 0 0.591 0 9 1.132 LNBVA 12.899 18.162 18.083 23.106 1.474 LFAGE 0 1.365 1.099 4.248 0.901 ODIR 0 4.866 5 14 2.072 TOBINQ 0.497 2.813 2.208 31.369 1.943 ROA -9.705-0.026 0.081 1.678 0.580 RETSD 0.010 0.036 0.035 0.149 0.012
Panel B: Summary statstcs of dependent varables mn mean medan max std. dev LEVER 0 0.203 0.107 1.063 0.231 DIVID 0 0.011 0.000 0.319 0.037 INVEST 0 0.192 0.102 1.856 0.242 NUMEST 0 0.834 0.693 3.526 0.842 FORERR 0 0.008 0.002 0.504 0.027 DISPERS 0 0.004 0.001 0.160 0.011 BIDASK 0 0.039 0.031 0.461 0.034 PVOL 10.118 16.181 16.242 21.070 1.592 TO 0.0002 0.145 0.106 1.128 0.125 Panel C: Correlatons between ndependent varables TSIZE TENURE TENHET PMBA PFTEAM PCPA CORE FCEO TSIZE 1 TENURE -0.1645 1 TENHET 0.1467 0.0578 1 PMBA -0.0153-0.0849-0.0166 1 PFTEAM -0.0590-0.4934-0.1095 0.0879 1 PCPA -0.1294 0.0253-0.0294 0.0903-0.0430 1 CORE -0.0225-0.1595-0.0019 0.1057 0.0686 0.0252 1 FCEO 0.1833 0.0524 0.0385-0.0707-0.0160-0.0495-0.1085 1 BOARDS 0.1677-0.0458-0.0004-0.0259 0.0443-0.0322-0.0515 0.0637 LNBVA 0.3493 0.1541-0.0007-0.1594-0.0793-0.1050-0.2515 0.3500 LFAGE 0.0615 0.3241 0.1077-0.0693-0.2358-0.0303-0.1764 0.1550 ODIR 0.1217-0.0637-0.0413-0.0432 0.0045-0.1017-0.1134 0.1900 TOBINQ -0.0193-0.1452 0.0474 0.0878 0.0481-0.0245 0.1674-0.1283 ROA 0.0499 0.1685 0.0584-0.0450-0.1328 0.0164-0.0748 0.1070 RETSD -0.1880-0.1869 0.0399 0.1444 0.1405 0.0122 0.2583-0.2507 BOARDS LNBVA LFAGE ODIR TOBINQ ROA RETSD BOARDS 1 LNBVA 0.1873 1 LFAGE 0.0725 0.2681 1 ODIR 0.1492 0.3422 0.1990 1 TOBINQ -0.0462-0.3202-0.2055-0.0928 1 ROA -0.0307 0.2857 0.1036-0.0454-0.0430 1 RETSD -0.0915-0.5710-0.1679-0.1784 0.2157-0.2576 1
Table 2 Selected statstcs related to a common factor analyss of eght measures of management qualty and reputaton. The sample conssts of 1,630 seasoned equty offerngs conducted between 1993 and 1997. TRF s the management team resources factor score obtaned usng common factor analyss on the frm-szeadjusted TSIZE, PMBA, PCPA, CORE, and PFTEAM. TSF s the management team structure factor score obtaned usng common factor analyss on the frm-age-adjusted TENURE, TENHET, and the frm-szeadjusted FCEO. Panel A: Estmated communaltes of eght management qualty measures TSIZE PCPA PFTEAM PMBA CORE TRF 0.0201 0.0234 0.0120 0.0186 0.0125 TENURE TENHET FCEO TSF 0.0018 0.0054 0.0036 Panel B: Egenvalues of the reduced correlaton matrx of eght management qualty measures 1 2 3 4 5 TRF 0.16768 0.13004 0.03907-0.07122-0.17893 TSF 0.07912-0.00073-0.06764 Panel C: Correlatons between the common factors and eght management qualty measures TSIZE PCPA PFTEAM PMBA CORE TRF 0.3645 0.1930 0.2487 0.7096 0.6341 TENURE TENHET FCEO TSF 0.4262 0.7312 0.5988 Panel D: Descrptve statstcs of the common factors extracted from eght management qualty measures TRF TSF Maxmum 1.371 1.550 Thrd quartle 0.203 0.155 Medan -0.022-0.009 Frst quartle -0.233-0.167 Mnmum -1.061-1.028 Mean 0.000 0.000
Table 3 Tests of dfferences n mean and medan fnancal and nvestment polcy varables for frms wth above and below medan management qualty and reputaton factor scores. TRF s the management team resources factor score obtaned usng common factor analyss on TSIZE, PMBA, PCPA, CORE, and PFTEAM. TSF s the management team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. LEVER s the long-term debt plus debt n current labltes over the long-term debt plus debt n current labltes plus the book value of common equty. DIVID s the sum of common and preferred dvdends over the earnngs before deprecaton, nterest, and taxes. INVEST s the rato of captal expendtures to the book value of assets. Medans are reported n brackets. The results of t-tests for the dfference n means and non-parametrc Wlcoxon sgned rank tests for the dfference n medans are reported n parenthess. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. below medan TRF above medan dfference below medan TSF above medan dfference LEVER 0.232 0.194-0.038 0.225 0.202-0.023 (-2.84)*** (-1.74)** [0.137] [0.092] [-0.045] [0.115] [0.098] [-0.017] (-2.39)** (-1.43) DIVID 0.019 0.006-0.013 0.005 0.020 0.015 (-1.14) (1.35) [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] (-3.01)*** (2.40)*** INVEST 0.221 0.234 0.013 0.227 0.227 0.000 (0.48) (0.01) [0.101] [0.105] [0.004] [0.102] [0.104] [0.002] (2.02)** (0.06)
Table 4 Relatonshp between leverage, dvdend payout rato, and management qualty and reputaton. LEVER s the rato of the long-term debt plus debt n current labltes to the long-term debt plus debt n current labltes plus the book value of common equty. DIVID s the rato of the sum of common and preferred dvdends to the earnngs before deprecaton, nterest, and taxes. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. All regressons nclude two-dgt SIC code ndustry dummes. Both dependent varables are wnsorzed at the 99 th percentle. Only observatons wth postve LEVER and DIVID are ncluded n the regressons. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 4 5 6 7 8 Dependent LEVER LEVER LEVER LEVER DIVID DIVID DIVID DIVID varable Intercept -0.933-0.981-0.937-1.038-0.050-0.054-0.049-0.060 (-8.69)*** (-9.60)*** (-8.97)*** (-9.92)*** (-2.44)** (-2.80)*** (-2.47)** (-3.07)*** TSIZE -0.003 (-1.06) -0.001 (-1.28) TENURE -0.002 (-1.03) 0.001 (1.19) TENHET -0.010 (-0.61) -0.002 (-0.59) PMBA -0.052 (-2.01)** -0.057 (-2.20)** 0.002 (0.37) PFTEAM -0.001-0.009-0.011 (-0.05) (-2.23)** (-3.03)*** PCPA -0.013 (-0.39) -0.001 (-0.19) CORE -0.076 (-3.16)*** -0.081 (-3.34)*** -0.008 (-2.06)** -0.010 (-2.33)** FCEO 0.010 (0.90) -0.003 (-1.24) BOARDS -0.002 (-0.47) -0.002 (-0.40) 0.002 (1.21) 0.001 (1.04) LNBVA 0.076 (14.64)*** 0.075 (15.67)*** 0.074 (15.61)*** 0.078 (15.97)*** 0.003 (2.16)** 0.002 (2.12)** 0.002 (1.95)* 0.002 (2.19)** LFAGE -0.016 (-2.42)** -0.016 (-2.70)*** -0.018 (-2.98)*** -0.018 (-2.97)*** 0.002 (1.44) 0.003 (1.91)* 0.003 (2.23)** 0.003 (2.10)** ODIR -0.001 0.000 0.000-0.000 0.002 0.002 0.002 0.002 (-0.29) (0.17) (0.14) (-0.05) (3.43)*** (3.20)*** (3.05)*** (3.02)*** TOBINQ -0.021 (-4.75)*** -0.022 (-4.79)*** -0.022 (-4.75)*** -0.022 (-4.77)*** 0.000 (0.35) -0.000 (-0.12) 0.000 (0.11) 0.000 (0.06) ROA -0.051 (-4.34)*** -0.052 (-4.30)*** -0.052 (-4.44)*** -0.052 (-4.39)*** 0.001 (0.53) 0.001 (0.59) 0.001 (1.10) 0.002 (1.18) RETSD -0.208 (-0.41) -0.315 (-0.63) -0.153 (-0.31) -0.262 (-0.52) TRF -0.044 (-3.50)*** -0.007 (-2.68)** TSF -0.010 (-0.54) -0.000 (-0.06) R 2 0.4485 0.4416 0.4448 0.4465 0.1452 0.1359 0.1319 0.1346 N 1528 1529 1529 1528 1504 1505 1505 1504
Table 5 Relatonshp between nvestment level and management qualty and reputaton. INVEST s the rato of captal expendtures to the book value of assets. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. All regressons nclude two-dgt SIC code ndustry dummes. Dependent varable s wnsorzed at the 99 th percentle. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 4 5 Dependent INVEST INVEST INVEST INVEST INVEST varable Intercept 0.618 (4.59)*** 0.705 (4.98)*** 0.633 (4.59)*** 0.605 (4.35)*** 0.703 (5.40)*** TSIZE 0.005 (2.16)** 0.006 (2.51)** TENURE -0.003 (-1.50) TENHET -0.002 (-0.08) PMBA 0.017 (0.39) PFTEAM 0.016 (0.76) PCPA 0.143 (2.98)*** 0.135 (2.87)*** CORE 0.054 (2.03)** 0.067 (2.51)** FCEO -0.009 (-0.60) BOARDS 0.004 (0.73) 0.005 (0.89) LNBVA -0.024 (-4.47)*** -0.028 (-5.21)*** -0.021 (-4.36)*** -0.021 (-4.31)*** -0.026 (-5.13)*** LFAGE -0.025 (-4.19)*** -0.030 (-5.29)*** -0.031 (-5.44)*** -0.028 (-5.02)*** -0.027 (-4.85)*** ODIR -0.003 (-0.91) -0.003 (-0.97) -0.002 (-0.86) -0.003 (-0.97) -0.003 (-0.93) TOBINQ 0.001 (0.39) 0.002 (0.49) 0.003 (0.71) 0.002 (0.51) 0.002 (0.46) ROA -0.041 (-1.84)* -0.041 (-1.88)* -0.044 (-2.03)** -0.044 (-1.96)* -0.043 (-1.89)* TRF 0.053 (3.30)*** TSF -0.020 (-0.89) R 2 0.3283 0.3171 0.3181 0.3167 0.3228 N 1518 1519 1519 1519 1518
Table 6 Seemngly unrelated regressons of leverage, dvdend payout rato, and nvestment level on management qualty and reputaton and other control varables. LEVER s the rato of the long-term debt plus debt n current labltes to the long-term debt plus debt n current labltes plus the book value of common equty. DIVID s the rato of the sum of common and preferred dvdends to the earnngs before deprecaton, nterest, and taxes. INVEST s the rato of captal expendtures to the book value of assets. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. All regressons nclude two-dgt SIC code ndustry dummes. LEVER, DIVID, and INVEST are wnsorzed at the 99 th percentle. Only observatons wth postve LEVER and DIVID are ncluded n the regressons. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 1 2 3 1 2 3 Dependent LEVER TRF TSF DIVID TRF TSF INVEST TRF TSF varable Intercept -1.062 (-8.22)*** -0.849 (-3.57)*** -0.142 (-0.79) -0.063 (-2.69)*** -0.520 (-2.23)** -0.172 (-0.96) 0.724 (5.28)*** -0.686 (-2.91)*** -0.131 (-0.72) BOARDS -0.002 (-0.45) 0.001 (1.60) 0.005 (0.99) LNBVA 0.080 (18.35)*** 0.062 (7.39)*** 0.004 (0.67) 0.003 (3.29)*** 0.036 (4.73)*** 0.006 (1.03) -0.027 (-6.06)*** 0.042 (5.49)*** 0.005 (0.87) LFAGE -0.020 (-3.79)*** -0.066 (-6.28)*** 0.016 (2.05)** 0.003 (2.65)*** -0.055 (-5.15)*** 0.010 (1.26) -0.024 (-3.80)*** -0.051 (-4.82)*** 0.011 (1.32) ODIR -0.000 (-0.16) -0.004 (-0.90) -0.007 (-1.95)* 0.002 (3.45)*** -0.001 (-0.25) -0.007 (-2.00)** -0.003 (-0.96) -0.003 (-0.58) -0.009 (-2.42)** TOBINQ -0.021 (-8.47)*** 0.002 (0.31) -0.001 (-0.20) 0.000 (0.16) 0.009 (1.75)* 0.001 (0.28) 0.001 (0.45) 0.008 (1.71)* 0.000 (0.03) ROA -0.052 (-5.81)*** -0.025 (-1.40) 0.036 (2.70)*** 0.001 (0.81) -0.006 (-0.31) 0.036 (2.56)** -0.042 (-4.45)*** 0.013 (0.82) 0.020 (1.56) LEVER -0.340 (-6.82)*** -0.027 (-0.71) DIVID -1.293 (-5.06)*** 0.027 (0.14) INVEST 0.311 (7.17)*** -0.089 (-2.64)*** RETSD -0.258 (-0.56) TRF -0.088 (-6.79)*** -0.013 (-5.09)*** 0.104 (6.97)*** TSF -0.020 (-1.15) -0.001 (-0.17) -0.040 (-2.08)** R 2 0.4422 0.1359 0.0929 0.1309 0.1313 0.0856 0.3168 0.1296 0.0881 N 1528 1528 1528 1504 1504 1504 1518 1518 1518
Table 7 Two-stage OLS regressons of leverage, dvdend payout rato, and nvestment level on management qualty and reputaton and other control varables. LEVER s the rato of the long-term debt plus debt n current labltes to the long-term debt plus debt n current labltes plus the book value of common equty. DIVID s the rato of the sum of common and preferred dvdends to the earnngs before deprecaton, nterest, and taxes. INVEST s the rato of captal expendtures to the book value of assets. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. REP s the natural logarthm of one plus the lead underwrter s reputaton as measured by Loughran and Rtter (2004). RESSD s the resdual standard devaton estmated usng market model over 255 tradng days for the fscal year of the ssue. All regressons nclude two-dgt SIC code ndustry dummes. LEVER, DIVID, and INVEST are wnsorzed at the 99 th percentle. Only observatons wth postve LEVER and DIVID are ncluded n the regressons. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 1 2 3 1 2 3 Dependent LEVER TRF TSF DIVID TRF TSF INVEST TRF TSF varable Intercept -1.421 (-4.97)*** 0.007 (0.81) 0.002 (0.01) -0.051 (-0.93) -0.756 (-2.92)*** -0.069 (-0.34) 1.199 (2.64)*** -0.763 (-2.94)*** -0.068 (-0.34) BOARDS -0.007 (-0.69) 0.029 (3.15)*** -0.007 (-1.14) 0.003 (1.59) 0.008 (0.93) -0.006 (-1.00) 0.013 (0.96) 0.008 (0.94) -0.007 (-1.11) LNBVA 0.098 (9.06)*** -0.055 (-5.12)*** -0.005 (-0.65) 0.003 (1.28) 0.025 (2.68)*** 0.001 (0.18) -0.051 (-2.69)*** 0.026 (2.77)*** 0.003 (0.40) LFAGE -0.024 (-1.14) -0.007 (-1.48) 0.018 (2.16)** -0.002 (-0.65) -0.055 (-5.08)*** 0.011 (1.33) -0.029 (-1.22) -0.055 (-5.17)*** 0.014 (1.65)* ODIR -0.010 (-1.37) 0.007 (1.38) -0.007 (-1.93)* 0.003 (1.95)* -0.007 (-1.33) -0.007 (-1.89)* 0.012 (0.98) -0.007 (-1.36) -0.008 (-2.16)** TOBINQ -0.018 (-3.10)** -0.006 (-0.32) -0.001 (-0.24) 0.000 (0.31) 0.007 (1.32) 0.001 (0.16) -0.001 (-0.20) 0.007 (1.41) -0.000 (-0.11) ROA -0.011 (-0.34) 0.176 (5.03)*** 0.034 (2.53)** -0.005 (-0.80) -0.004 (-0.24) 0.032 (2.26)** -0.079 (-2.26)** 0.002 (0.12) 0.021 (1.62) REP 3.301 (3.52)*** 0.033 (1.23) 0.179 (5.03)*** 0.017 (0.61) 0.179 (5.03)*** 0.005 (0.19) RESSD -0.824 (-3.20)*** -1.022 (-1.42) 3.176 (3.35)*** -1.016 (-1.38) 2.967 (3.24)*** -1.164 (-1.64) TRF -0.479 (-2.55)** -0.062 (-1.95)* 0.397 (1.50) TSF -1.182 (-1.59) 0.180 (1.17) 1.597 (1.37) R 2-1.5567 0.1563 0.0960-1.8865 0.1520 0.0880-2.8482 0.1493 0.0909 N 1528 1528 1528 1501 1501 1501 1515 1515 1515
Table 8 Tests of dfferences n mean and medan asymmetrc nformaton and lqudty varables for frms wth above and below medan management qualty factor scores. TRF s the management team resources factor score obtaned usng common factor analyss on TSIZE, PMBA, PCPA, CORE, and PFTEAM. TSF s the management team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. NUMEST s the natural logarthm of one plus the number of analysts followng the frm at the end of the fscal year precedng the ssue. FORERR s the forecast error, defned as the rato of the absolute dfference between analysts mean EPS estmate and actual EPS to the stock prce. DISPERS s the rato of the standard devaton of analysts forecasts to the stock prce. BIDASK s the medan monthly bd-ask spread as a percentage of the stock prce, calculated over the precedng fscal year. PVOL s the natural logarthm of the product of the stock prce and the monthly volume of trade, averaged over the fscal year precedng the offer. TO s the monthly tradng volume dvded by the number of shares outstandng, averaged over the fscal year precedng the offer. Medans are reported n brackets. The results of t-tests for the dfference n means and non-parametrc Wlcoxon sgned rank tests for the dfference n medans are reported n parenthess. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. below medan TRF above medan dfference below medan TSF above medan dfference NUMEST 0.723 0.945 0.222 0.785 0.884 0.099 (5.36)*** (2.39)*** [0.693] [1.099] [0.406] [0.693] [1.099] [0.406] (5.32)*** (2.50)** FORERR 0.010 0.006-0.004 0.008 0.007-0.001 (-2.05)** (-0.78) [0.002] [0.002] [0.000] [0.002] [0.002] [-0.000] (0.96) (-0.80) DISPERS 0.004 0.004 0.000 0.005 0.003-0.002 (0.55) (-2.18)** [0.001] [0.001] [0.000] [0.001] [0.001] [-0.000] (1.52) (-1.66)* BIDASK 0.042 0.037-0.005 0.040 0.038-0.002 (-2.73)*** (-1.55)* [0.032] [0.031] [-0.001] [0.032] [0.031] [-0.001] (-2.26)** (-1.51) PVOL 15.919 16.436 0.517 16.171 16.196 0.025 (6.25)*** (0.30) [15.954] [16.478] [0.524] [16.243] [16.230] [-0.013] (6.16)*** (0.22) TO 0.126 0.163 0.037 0.148 0.141-0.007 (5.76)*** (-1.05) [0.095] [0.123] [0.028] [0.113] [0.104] [-0.009] (5.51)*** (-0.51)
Table 9 Panel A. Relatonshp between the level of asymmetrc nformaton and management qualty and reputaton. NUMEST s the natural logarthm of one plus the number of analysts followng the frm at the end of the fscal year precedng the ssue. FORERR s the rato of the absolute dfference between analysts mean EPS estmate and actual EPS to the stock prce. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 4 5 6 7 8 Dependent NUMEST NUMEST NUMEST NUMEST NUMEST FORERR FORERR FORERR varable Intercept -3.297-2.694-3.561-2.765-2.734 0.046 0.037 0.034 (-11.88)*** (-10.59)*** (-13.50)*** (-10.86)*** (-10.86)*** (1.92)* (2.08)** (1.84)* TSIZE 0.037 (4.09)*** 0.040 (4.66)*** -0.000 (-0.22) TENURE 0.013 (1.69)* -0.001 (-2.00)** TENHET 0.033 (0.53) -0.002 (-0.79) PMBA 0.048 (0.41) -0.005 (-1.47) PFTEAM -0.032 (-0.42) -0.005 (-1.13) PCPA 0.026 (0.18) -0.014 (-2.41)** -0.014 (-2.51)** CORE 0.629 (6.60)*** 0.641 (6.84)*** -0.002 (-0.70) FCEO 0.028 (0.58) -0.001 (-0.41) BOARDS 0.049 (2.52)** 0.052 (2.65)*** 0.050 (2.63)*** -0.000 (-0.47) -0.000 (-0.51) LNBVA 0.181 0.174 0.215 0.191 0.188-0.002-0.002-0.002 (10.52)*** (10.87)*** (14.36)*** (12.59)*** (12.49)*** (-1.73)* (-1.96)* (-1.75)* LFAGE 0.133 (5.48)*** 0.135 (5.95)*** 0.148 (6.52)*** 0.131 (5.73)*** 0.145 (6.40)*** 0.002 (1.72)* 0.002 (1.85)* 0.002 (1.75)* ODIR -0.012 (-1.14) -0.014 (-1.38) -0.011 (-1.13) -0.016 (-1.54) -0.014 (-1.38) 0.000 (0.72) 0.001 (0.88) 0.001 (0.96) TRF 0.296 (5.87)*** -0.003 (-1.36) TSF 0.150 (2.06)** -0.004 (-1.11) R 2 0.1954 0.1695 0.1789 0.1613 0.1781 0.0171 0.0125 0.0116 N 1629 1630 1630 1630 1629 915 915 915
Panel B. Relatonshp between the level of asymmetrc nformaton and management qualty and reputaton. DISPERS s the rato of the standard devaton of analysts forecasts to the stock prce. BIDASK s the medan monthly bd-ask spread as a percentage of the stock prce, calculated over the precedng fscal year. See Table 1 for defntons of ndependent varables. TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 4 5 6 7 8 Dependent DISPERS DISPERS DISPERS DISPERS BIDASK BIDASK BIDASK BIDASK varable Intercept 0.017 0.019 0.021 0.018 0.211 0.203 0.213 0.206 (2.44)** (2.95)*** (3.21)*** (2.90)*** (9.75)*** (11.39)*** (11.80)*** (11.86)*** TSIZE 0.000 (0.34) -0.001 (-4.22)*** -0.001 (-4.75)*** TENURE -0.000 (-2.31)** -0.000 (-2.95)*** 0.000 (0.53) TENHET -0.001 (-1.32) -0.004 (-1.06) PMBA 0.004 (1.33) -0.004 (-0.74) PFTEAM -0.001 (-0.59) 0.001 (0.34) PCPA -0.006 (-3.01)*** -0.006 (-3.38)*** -0.015 (-2.25)** -0.012 (-1.90)* CORE 0.003 (1.67)* -0.004 (-0.89) FCEO -0.000 (-0.07) -0.002 (-0.66) BOARDS -0.000 (-0.48) -0.000 (-0.61) -0.001 (-1.08) -0.001 (-1.45) LNBVA -0.001-0.001-0.001-0.001-0.010-0.010-0.010-0.010 (-2.34)** (-2.97)*** (-3.20)*** (-3.03)*** (-7.76)*** (-8.41)*** (-9.50)*** (-9.47)*** LFAGE 0.002 (3.60)*** 0.002 (3.59)*** 0.001 (3.24)*** 0.002 (3.49)*** 0.003 (2.60)*** 0.003 (2.97)*** 0.003 (3.06)*** 0.003 (2.72)*** ODIR 0.001 (2.99)*** 0.001 (3.18)*** 0.001 (3.21)*** 0.001 (3.18)*** 0.001 (1.53) 0.001 (1.62) 0.001 (1.32) 0.001 (1.32) TRF 0.002 (2.10)** -0.007 (-3.04)*** TSF -0.002 (-1.93)* -0.006 (-1.31) R 2 0.0570 0.0476 0.0463 0.0493 0.1582 0.1550 0.1464 0.1492 N 786 786 786 786 1399 1400 1400 1399
Table 10 Relatonshp between lqudty and management qualty and reputaton. PVOL s the natural logarthm of the product of the stock prce and the monthly tradng volume, averaged over the fscal year pror to the offer. TO s the rato of the monthly tradng volume to the number of shares outstandng, averaged over the fscal year pror to the offer. See Table 1 for defntons of ndependent varables. SHAROUT s the number of shares outstandng at the end of the fscal year pror to the offer (n mllons). TRF s the team resources factor score obtaned usng common factor analyss on the TSIZE, PMBA, PCPA, PFTEAM, and CORE. TSF s the team structure factor score obtaned usng common factor analyss on TENURE, TENHET, and FCEO. ***, **, and * ndcate sgnfcance at the 1, 5, and 10 percent levels, respectvely. 1 2 3 4 5 6 7 8 9 10 11 12 Dependent PVOL PVOL PVOL PVOL PVOL PVOL PVOL TO TO TO TO TO varable Intercept 6.332 7.744 7.191 7.114 7.211 6.308 7.328 0.156 0.277 0.237 0.168 0.246 (10.78)*** (14.82)*** (13.70)*** (13.33)*** (13.71)*** (11.05)*** (14.11)*** (3.19)*** (7.06)*** (5.88)*** (3.95)*** (6.40)*** TSIZE 0.065 (4.17)*** 0.081 (5.65)*** 0.004 (3.42)*** 0.005 (4.38)*** TENURE -0.035 (-2.66)*** -0.002 (-1.80)* TENHET 0.203 (1.76)* 0.283 (2.46)** 0.014 (1.29) PMBA 0.384 (1.75)* 0.544 (2.45)** 0.009 (0.42) PFTEAM -0.017 (-0.12) 0.017 (1.36) 0.022 (2.06)** PCPA 0.646 (2.65)*** 0.520 (2.14)** 0.016 (0.75) CORE 0.707 (4.05)*** 0.831 (4.74)*** 0.065 (4.05)*** 0.073 (4.54)*** FCEO -0.030 (-0.34) -0.006 (-0.71) BOARDS 0.034 (1.19) 0.044 (1.53) -0.003 (-0.96) -0.002 (-0.67) LNBVA 0.533 (15.55)*** 0.475 (15.23)*** 0.524 (17.14)*** 0.537 (17.41)*** 0.531 (17.38)*** 0.555 (17.87)*** 0.522 (17.31)*** -0.000 (-0.11) -0.005 (-2.12)** -0.001 (-0.61) 0.001 (0.28) -0.001 (-0.69) LFAGE -0.479-0.543-0.565-0.548-0.551-0.522-0.504-0.027-0.032-0.031-0.03-0.029 (-10.15)*** (-12.90)*** (-13.23)*** (-12.90)*** (-13.03)*** (-12.17)*** (-11.59)*** (-6.40)*** (-8.54)*** (-7.87)*** (-7.82)*** (-7.50)*** ODIR -0.043-0.039-0.033-0.037-0.034-0.034-0.037-0.006-0.006-0.006-0.006-0.006 (-2.40)** (-2.26)** (-1.85)* (-2.12)** (-1.94)* (-1.93)* (-2.14)** (-4.27)*** (-4.49)*** (-4.49)*** (-4.21)*** (-4.10)*** SHAROUT 0.014 (6.57)*** 0.014 (6.63)*** 0.014 (6.17)*** 0.014 (6.12)*** 0.014 (6.11)*** 0.014 (6.05)*** 0.013 (6.28)*** TRF 0.619 (6.87)*** 0.043 (5.00)*** TSF 0.046 (0.33) -0.002 (-0.19) R 2 0.3794 0.3630 0.3495 0.3505 0.3498 0.3583 0.3665 0.0969 0.0796 0.0717 0.0819 0.0840 N 1445 1446 1445 1446 1446 1446 1445 1446 1447 1447 1447 1446
Fgure 1 Model of relatonshp between management reputaton and nformaton asymmetry facng the frm n the equty market. Frm s fnancal polcy s chosen and observed by nvestors; Investors value frm s equty (Step 2) True value of frm s cash flow s realzed and observed by equty market nvestors; Investors update manageral reputaton for honesty (Step 3) Steps 1, 2, and 3 occur for new frm t = 0 t = 1 t = 2 Manager announces frm s future cash flow to equty market nvestors (Step 1) Manager leaves current frm and jons new frm End of game Perod one Perod two
Fgure 2 Relatonshp between management qualty and nvestment. As management qualty ncreases from low (L) to hgh (H), the scale of the frm s nvestment ncreases from I * L to I * H. Net Present Value of Last Dollar Invested Hgh (H) Management Qualty Low (L) Management Qualty Increasng Management Qualty I * L I * H Equlbrum Scale Expands wth Management Qualty Scale of Investment