IPO PERFORMANCE UNDER LOW INFORMATION ASYMMETRY AND LOW AGENCY CONFLICTS: THE CASE OF DEMUTUALIZED INSURERS

Transcription

1 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 IPO PERFORMANCE UNDER LOW INFORMATION ASYMMETRY AND LOW AGENCY CONFLICTS: THE CASE OF DEMUTUALIZED INSURERS Kenneh A. Kim*, Piman Limpaphayom**, Shanhong Wu*** Absrac One of he hypoheses pu forh o explain he pos-issue underperformance of IPO socks is ha managers ake advanage of windows of opporuniy by iming he offerings when shares are overvalued. This sudy invesigaes he hypohesis by examining he pos-ipo performance of insurance firms ha wen public afer going hrough a process called demuualizaion, a srucural conversion from muual o sock company. Because he demuualizaion process is highly regulaed, managers of hese insurers have lile discreion o ime he issue. Furhermore, he demuualizaion process is very ransparen so hese IPOs should have lower informaion asymmery. Empirical resuls show ha, on average, demuualized insurance IPOs do no exhibi poor performance compare o various benchmarks. The resuls yield indirec suppor o he noion ha managers of oher ypes of IPO firms ake advanage of window of opporuniy which, in urn, leads o poor long-run performance. * Sae Universiy of New York a Buffalo **Corresponding Auhor: Sasin Graduae Insiue of Business Adminisraion of Chulalongkorn Universiy, Chula 12, Phyahai Road, Bangkok 10330, Thailand. Telephone: (662) piman.limpaphayom@sasin.edu *** Sae Universiy of New York a Buffalo The auhors are indebed o Gene C. Lai, Michael J. McNamara, and Jay Rier for commens on earlier drafs of his paper. The firs auhor graefully acknowledges financial suppor from a SUNY-Buffalo SOM Small Gran Fund. All remaining errors are our own. I. Inroducion I is well-documened in he lieraure ha iniial public offerings (IPOs) underperform in he longrun. Numerous sudies have documened significan declines in marke performance (Ibboson, 1975; Rier, 1991; Loughran and Rier, 1995) and operaing performance (Jain and Kini, 1994) of IPO socks. There is also a ho issue marke phenomenon in which firms in cerain indusries go public in a paricular high-volume period exhibiing worse performance han hose making offerings in low-volume years (Rier, 1984). One explanaion is relaed o he poenial agency conflics when a privae firm urns public (Jain and Kini, 1994). Jensen and Meckling (1976) posi ha, wih reducion in shareholding, managers may no longer ac on behalf of all shareholders. For insance, i is hypohesized ha, due o agency conflics, managers will ake advanage of ouside invesors by iming he issues when shares are overvalued (Loughran and Rier, 1995). Furhermore, managers ry o expropriae wealh from ouside invesors by using IPO proceeds o inves in nonvalue maximizing aciviies which, in urn, resuls in poor operaing performance. Moreover, invesors end o overesimae he share prices due o he informaion asymmery beween managers and invesors. In he end, i is concluded ha he underperformance of IPO socks are a resuls of over-opimism by invesors and managers aking advanage of his window of opporuniy (Rier, 1991). I is possible o es he hypohesis ha informaion asymmery and agency conflics lead o poor pos-ipo performance indirecly by examining a unique group of companies ha wen public under an environmen which has relaively low informaion asymmery and lower agency conflics. Recenly, a grea deal of aenion focuses on 525

2 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 sudying muual insurance companies ha conver o sock insurance companies in a process known as demuualizaion (e.g., McNamara and Rhee, 1992; Viswanahan and Cummins, 2003; Carson, Forser and McNamara, 1998). A muual insurance company is lierally owned by is policyholders, while a sock insurance company is owned by ouside invesors. For decades, boh organizaional forms have co-exised in he insurance indusry, as each organizaional form has is own se of comparaive advanages (Fama and Jensen, 1983). However, wih he recen wave of demuualizaions in he insurance indusry, financial economiss have become inrigued by he underlying moive ha is driving hese conversions, and wheher or no he moive is jusified, as a demuualizaion is a dramaic, cosly, rigid, and ime-consuming process. The objecive of his sudy is o idenify a se of demuualized (boh life and properycasualy) insurance firms ha subsequenly conduc an iniial public offering (IPO) and assesses heir long-run performance pos-ipo. There are several reasons ha convering muual insurance firms have relaively lower informaion asymmery and lower agency conflics. Firs of all, he process of demuualizaion is quie arduous, ime-consuming, and ransparen wih close monioring by regulaors. The saes also require ha companies hold public hearings o inform general public and invesors. Because of he sric and relaively ransparen demuualizaion process, convering insurance firms ha wen public may acually exhibi relaively lower informaion asymmery compared o IPOs from oher indusries. In addiion, he lieraure has shown ha he muual srucure conrols agency conflics beween managers and owners less effecively han does he sock srucure. Given ha demuualized insurance firms separae he roles of policyholders and owners, demuualized insurance firms should experience lower agency coss relaed o monioring. The reason is ha ouside monioring (i.e., marke discipline) is more effecive in conrolling managerial discreion han monioring by policyholders (Mayers and Smih, 1981). Consequenly, i is unlikely ha managers of demuualized insurance IPOs have incenives o expropriae wealh from ouside shareholders (i.e., relaively lower agency conflics han IPOs from oher indusries). Furher, he primary reason being cied for demuualizaion is access o capial Viswanahan and Cummins, 2003). Because muual insurers can increase capial by selling more policies or use surplus noes, hey have limied capial raising capabiliies. Wih increased compeiion from banks and oher financial insiuions, finding addiional capial has become even more difficul for muual insurers. Consequenly, i can be argued ha he iniial public offerings by muual insurers are moivaed by a genuine economic reason, no agency-relaed incenives. If insurance firms genuinely demuualize o gain access o capial, hen hese insurance IPOs should no underperform in he long-run. This conenion is wha being empirically esed in he paper. In sum, examining pos-issue performance of convering insurance firms should provide insighs, ex pos, o he moive behind he demuualizaion and subsequen public offerings and he sources of organizaional efficiencies resuling from demuualizaion. Finally, he findings should also shed ligh on IPO performance when informaion asymmery is relaively low and agency coss are reduced. The null hypohesis is ha demuualized firms IPOs do no underperform oher IPOs and he benchmarks in he long-run. The firs alernaive hypohesis is ha, like IPOs from oher indusries, demuualized insurance IPOs exhibi underperformance in he long-run. The oher alernaive hypohesis pu forh in he insurance lieraure is ha, because demuualizaion enhances efficiency, demuualized IPOs ouperform he benchmarks in he long-run (Lai, McNamara and Yu, 2007; Viswanahan, 2006). In conras o he findings from previous sudies in he insurance lieraure, he empirical findings from his sudy do no lead o he rejecion of he null hypohesis. I is documened ha demuualized insurers ha immediaely issued IPOs exhibi a normal marke performance when compared o an insurance secor benchmark and mached-firm benchmarks. Furhermore, oher ypes of insurance IPOs (e.g., sock insurance firms making public offerings) sill exhibi poor pos-issue marke performance. I is worh menioning ha hese resuls provide indirec suppor he window of opporuniy hypohesis for why IPOs underperform in he long-run. This conclusion is suppored by he fac ha insurance firms ha have he abiliy o ime heir issues (e.g., sock firms) sill exhibi underperformance like ha observed among IPOs from oher indusries. A comparison of pos-ipo operaing performance beween demuualized insurance firms and oher insurance firms confirms he sock reurns findings. In addiion, demuualized insurance IPO firms also show relaively high marke-o-book (M/B) raio, price-o-earnings (P/E) raio, and earnings per share (EPS). Taken ogeher, he overall findings imply ha demuualizaion decisions and subsequen public offerings are economically moivaed and jusified ex pos. The resuls sugges ha IPOs do no underperform in he long-run when, a he ime of he issue, informaion asymmery is low and agency coss are reduced. The res of he paper begins by describing he lieraure on pos-ipo performance. Nex, he benefis and drawbacks of he muual organizaional form and he sock organizaional form are summarized. Hypoheses concerning he 526

3 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 pos-issue marke performance and operaing performance of demuualized insurance IPOs are formulaed. The nex secion discusses he sample characerisics and presens he empirical findings. A conclusion is offered a he end. II. Pos-issue performance of IPOs The reasons for going public involve he rade-offs beween he benefis of being publicly raded and he associaed coss (Chemmanur and Fulghieri, 1999). Financial economiss have proposed several benefis of going public. For he enrepreneurs, hey gain from having a more diversified porfolio (Benninga, Helmanel and Sarig, 2005). Furhermore, Holmsröm and Tirole (1993) and Bolon and von Thadden (1998) conend ha increased monioring by ousiders and increased liquidiy could posiively affec firm value. Having shares sold publicly also faciliaes firm valuaion by invesors (Benvenise and Spind, 1989; Dow and Goron, 1997; Subrahmanyam and Timan, 1999) who, in urn, can use he marke price informaion o make fuure invesmen and compensaion decisions. However, here are also numerous coss of going public o he original owners. They have o give up conrol and increase disclosure of inside informaion o ousiders which, in urn, can reduce he firm s compeiive advanage. More imporanly, here is also a cos of separaing ownership and conrol (i.e., he agency cos of equiy) (Jensen and Meckling, 1976). The agency cos of equiy, along wih informaion asymmery, poenially leads o a siuaion in which enrepreneurs aemp o expropriae wealh from new ousider shareholders. This expropriaion of wealh can lead o high levels of underpricing a he iniial public offering and poor long-run performance. There is empirical evidence ha firms ime he decisions o go public (Rier, 1984). Rier (1991) finds ha IPO firms during exhibi poor marke performance agains maching firms for hree years afer iniial public offerings. Loughran and Rier (1995) furher documen a poor long-run underperformance for a 5-year horizon. Jain and Kini (1994) documen a significan decline in operaing performance afer iniial public offerings of firms ha wen public during The decline in pos-issue invesmen levels is also documened world-wide by Loughran, Rier and Rydqvis (1994) and Pagano, Panea and Zingales (1998). Rier (1991) concludes ha invesors are ofen oo opimisic abou he poenial of young firms and ha companies ake advanage of hese windows of opporuniy. Jain and Kini (1994) conend ha poor operaing performance is a funcion of informaion asymmery and agency conflics. For insance, managers ry o manipulae accouning numbers prior o public offerings or decide o go public during a period of unusually high performance ha canno be susained. Consequenly, IPOs end o exhibi poor pos-issue operaing performance. In conclusion, he poor pos-issue performance of IPOs has been exensively documened by previous research. This phenomenon is aribuable o he high degrees of agency conflics and informaion asymmery beween he firm s owners/managers and ouside invesors. In he end, i is conjecured ha firms ake advanage of he window of opporuniy by issuing shares when hey are, on average, over priced. Alhough he conjecure is quie plausible, i is fairly difficul o empirically and direcly es his conjecure. One approach is o examine long-run performance of a unique group of firms ha wen public in an environmen ha has relaively low informaion asymmery and agency conflics. A comparison of long-run performance of his group of firms should yield addiional insighs on he conjecure. The nex secion describes a small and unique group of insurance companies ha wen public afer going hrough a process called demuualizaion, a conversion from a muual form of organizaion o a sock form. Because of he unique characerisics of he demuualizaion process, hese IPOs allow he es of he conjecure on he poor long-run performance of IPOs. III. Organizaional srucure of insurance companies Wihin he insurance indusry, here are wo major ypes of organizaional srucures, muual and sock. The sock form is similar o a radiional public corporaion srucure in ha here is a separaion of ownership and conrol. Managers manage day-o-day aciviies whereas sockholders receive residual claims from operaions. Furher, shareholders and fixed-claim holders (i.e., policyholders/cusomers) are usually separae eniies in he sock form. In conras, in he muual organizaional form he policyholders (fixed-claim holders) have ownership righs o he firm which effecively merges he sockholder and cusomer funcions. Combining he cusomer and owner funcions helps alleviae any agency conflics beween shareholders who desire wealh maximizaion and cusomers who are relaively risk-averse. However, in he muual form, he agency cos beween managers and policyholders can be significan in he absence of marke discipline. The sock form essenially has differen drawbacks and benefis from he muual form. The sock form poenially creaes a conflic beween ouside invesors and policyholders, bu i enjoys he benefi of marke-based mechanisms ha 527

4 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 monior and conrol opporunisic managerial behavior. Consequenly, boh organizaional forms should be able o coexis even in a compeiive opimizing equilibrium environmen. Tha is, each insurance firm should adop he organizaional form ha is opimal given is unique characerisics. For insance, Mayers and Smih (1981, 1988) observe ha muual insurance companies end o focus on lines ha require relaively lower managerial discreion. A he same ime, sock insurers end o specialize in business lines ha require more managerial decision-making auhoriy. In oher words, he naure of he operaion dicaes he choice of organizaional srucure, hereby allowing boh forms o co-exis (Mayers and Smih, 1986; Smih, 1986). This co-exising hypohesis also explains he fac ha muual insurers end o focus on business lines wih longer duraions. This is because he incenives o ake excessive risk are miigaed when owners and fixed-claim holders are he same eniy (Mayers and Smih, 1981; Poier and Sommer, 1997). Cummins, Weiss, and Zi (1999) conend ha he choice of producion echnology also allows muual and sock insurers o opimize heir efficiency. Muual insurers can also make use of corporae governance mechanisms o alleviae problems beween owners and managers. For example, Mayers, Shivdasani, and Smih (1997) find ha muual insurers end o have a higher proporion of ouside direcors which, in urn, allow hem o effecively conrol agency conflics. In he end, each ype of organizaional form appears o operae in an environmen where i holds a comparaive advanage (Fama and Jensen, 1983). The recen wave of conversions, from he muual form o he sock form, implies ha he sock form has become more advanageous for muual insurers. The primary reason cied for demuualizaion is access o capial. The explanaion is ha, by design, muual insurers have limied capial raising capabiliies addiional capial can only come from an increase in heir cusomer base and/or reained earnings. Wih increased compeiion and a changing business environmen, muual insurers find i more difficul o raise addiional capial o finance heir operaions and invesmen. Examining insurance firm conversions during , Viswanahan and Cummins (2003) find ha he liquidiy consrain and he need o increase managerial discreion drive he demuualizaion process, consisen wih he access o capial hypohesis. In addiion, Viswanahan and Cummins (2003) also argue ha increased produc complexiy requires greaer managerial conrol ha only he sock form permis. This is consisen wih he noion ha he owners/cusomers of muual insurers may no be able o effecively monior he acion of muual managers (Mayers and Smih, 1981). Boose (1990) finds ha muual insurers end o have higher expenses han heir sock counerpars. Wells, Cox and Gaver (1995) also find ha, consisen wih Jensen (1986), muual insurers have relaively higher levels of free cash flows han sock insurers. Finally, Viswanahan and Cummins (2003) also posi ha muual managers may desire sock-based compensaion for heir services. The conversion process is quie lenghy, rigid, and cosly, as i involves he board of direcors, policyholders, and regulaors in a significan manner. Muual managers also lose flexibiliy under he new sock organizaional form as hey now face marke monioring. Therefore, he poenial benefis of demuualizaion mus ouweigh he conversion coss. Examining he conversions of muual savings and loans o sock charer, Masulis (1987) finds ha, on average, major claimans in he savings and loans conversions gain benefi from he demuualizaion. Cole and Mehran (1998) examine he performance of demuualized hrif insiuions and find ha marke performance of convering hrifs improves drasically. Several sudies have also documened operaing performance improvemens immediaely afer demuualizaion in he insurance indusry (Carson, Forser and McNamara, 1998; McNamara and Rhee, 1992). Wih he recen wave of iniial public offerings of insurers, i provides a unique opporuniy o examine pos-conversion marke and operaing performance from a differen (i.e., marke-based) perspecive. The nex secion discusses hypoheses on he long-run performance of iniial public offerings for insurers ha jus wen hough he demuualizaion process. IV. Hypohesis The null hypohesis is ha demuualized insurance firms IPOs will no exhibi underperformance as documened in IPOs from oher indusries. The compeing hypohesis in his sudy is ha demuualized insurance firms IPOs will perform as poorly as oher IPOs. In oher words, managers of demuualized insurers ake advanage of informaion asymmery and he window of opporuniy by going public when shares are overpriced which, in urn, resuls in poor long-run marke performance (Rier, 1991; Loughran and Rier, 1995). Alernaively, anoher compeing hypohesis is ha, because demuualizaion enhances efficiency, demuualized IPOs should exhibi superior pos-issue performance (Lai, McNamara and Yu, 2007; Viswanahan, 2006). In his sudy, i is conended ha here are heoreical reasons o rejec he null hypohesis. Firs of all, i can be argued ha insurance firms ha jus wen hrough demuualizaion should have relaively lower informaion asymmery han oher ypes of firms ha conduced iniial public 528

5 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 offerings. The reason is ha, because of sae regulaions, he demuualizaion process is very ime-consuming and ransparen (Viswanahan and Cummins, 2003). Iniially, he board of direcors mus approve he iniiaion of he demuualizaion plan. Once he approval is obained, he insurer mus provide deailed informaion o independen accouning and invesmen advisors who will work wih he insurer o draf he reorganizaion plan. The sae regulaor will also examine he demuualizaion plan in deail and acively paricipae in he process. As a par of he sae s requiremens, public hearings will also be held o gaher commens and opinions from he general public and policyholders. Evenually, a majoriy approval from he policyholders is required before he insurer can proceed wih he conversion. The process, which can ake up o wo years, is necessary o make sure ha policyholders are fairly compensaed for relinquishing heir ownership righs. Because relevan informaion has been revealed hrough ou he conversion process, he degrees of informaion asymmery for convering muual insurers IPOs should be alleviaed. I should also be noed ha mos of he demuualized firms are also relaively large and well-esablished companies (e.g., Prudenial, MeLife, and John Hancock) so he degrees of informaion asymmery should also be significanly lower han small, highgrowh IPOs in oher indusries. Secondly, i can be argued ha convering insurers should have relaively lower flexibiliy in iming he public offerings. The reason is ha he financial marke generally expecs public offerings immediaely afer he reorganizaion plan is approved regardless of he marke condiions. Because firms have lile conrol over he pace of he demuualizaion process, managers of convering insurers should have relaively lower abiliy o ime he marke or o ake advanage of any kind of windows of opporuniy. As a resul, i is inferred ha he demuualizaion process leads o relaively less opporuniy for wealh expropriaion by demuualized insurance managers han for managers of IPO firms in oher indusries. Thirdly, i can be argued ha muual insurers are going public for a genuine economic reason, no agency-relaed incenives. Previous sudies have pu forh access o capial as he main moivaion for insurance firms o conduc he conversion. The reason is ha muual insurers conver in order o mee wih he changing compeiive environmen in he insurance indusry. For he pas wo decades, insurance companies have been facing ough compeiion from banks (Carow, 2001) and oher financial insiuions (e.g., muual funds) which can now sell similar producs ha used o be sold exclusively by insurance companies. Therefore, muual insurers need o raise addiional capial o keep up wih inense compeiion. However, muual insurers can only raise funds hrough reained earnings, surplus noes or selling more policies which, in urn, hampers heir abiliy o compee. To mainain heir compeiiveness, he conversion from muual o sock form is a sensible course of acion for hese insurance firms. Because hese insurance IPOs are in fac moivaed by an economically jusified reason, i is more likely ha funds raised from IPOs will be used o inves in value-maximizing (posiive NPV) projecs. Jain and Kini (1994) posi ha managers of IPO firms generally have incenive o inves in non-valuemaximizing projecs which, in urn, leads o poor pos-issue operaing performance. However, hese insurance IPOs should no suffer poor pos-issue performance as observed in IPOs from oher indusries because of heir unique circumsances. Furhermore, i is hypohesized ha convering life insurers should benefi more from he public offerings as he compeiion from oher financial insiuions comes in he areas ha life insurance used o dominae (e.g., annuiies). Finally, here are also addiional exenuaing circumsances ha should provide suppor o he alernaive hypohesis. The reason is ha insurance firms ha conver heir organizaion srucure from muual o sock should experience significan decreases in agency conflics. Under he muual form, policyholders have very limied power or incenive o monior he managers. The reason is ha, for any owner/policyholder, aking acions agains he managemen will resul in a pro raa gain only for one share/policy. Consequenly, policyholder voing for muual insurance companies has been exremely low a less han one percen (Viswanahan and Cummings, 2003). For managers of sock insurance companies, he level of monioring is relaively higher han ha of muual insurers because shareholders can gain from heir acions. The end resul is ha managers of muual insurers do no always ac on behalf of he shareholders leading o non-value-maximizing invesmens or perquisie consumpion (Mayers and Smih, 1981). Previous sudies have documened ha muual insurers have relaively higher expenses and free cash flows han sock insurers (Boose, 1990; Wells, Cox and Gaver, 1995). In summary, i is concluded ha he exen of agency conflic should be reduced when insurers are ransiioning from muual o sock form of organizaion because ouside invesors can provide marke discipline and monioring of managerial discreion ha policyholder/owners canno provide effecively (Mayers and Smih, 1981). Examining insurance IPOs also provides a unique opporuniy o es he conjecure pu forh in he lieraure (e.g., Rier s) indirecly because here is anoher group of insurance IPOs ha were already sock firms before going public. Unlike convering insurance IPOs, hese sock insurance 529

6 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 firms have he abiliy o ime he marke and ake advanage of privae informaion. Furhermore, he informaion asymmery beween owners/managers and ouside invesors should be relaively higher for IPOs of sock insurers han ha for convering muual insurance IPOs. The reason is ha sock insurers only need o go hrough he process wih heir invesmen bankers wihou any requiremen o disseminae criical informaion o he public or ouside invesors. Consequenly, i is hypohesized ha insurance IPOs ha were already sock firms before going public should sill suffer from he usual agency conflic and informaion asymmery which, in urn, leads o poor pos-issue performance. I should be quie apparen ha sample and benchmark selecion are criical for he es. To be qualified ino he sample, he firms mus be posiively idenified as pure life or properycasualy insurance companies ha have jus gone hrough he demuualizaion process immediaely before going public. There are concurren sudies examining a larger sample of demuualized insurers IPO performance. However, hose sudies have several limiaions wih regard o esing he conjecure on IPO underperformance. For example, Viswanahan (2006) primarily examines only he insurance IPOs. She finds ha demuualized IPOs end o underprice more han non-demuualized IPOs and conends ha he high underpricing is consisen wih he legal liabiliy hypohesis while making no comparison wih IPOs from oher indusries. She also looks a long-run sock marke performance and finds ha demuualized insurance firms perform beer han he CRSP Index, which consiss of all NYSE, AMEX and NASDAQ firms. Lai, McNamara, and Yu (2007) also sudy shor-run and long-run sock reurns of demuualized IPO firms and show findings consisen wih Viswanahan (2006). In he end, boh sudies conclude ha demuualizaion enhances he wealh of he shareholders. However, boh sudies only compare long-run performance of insurance IPOs wih he marke-wide index (NYSE/AMEX/NASDAQ Index) so heir resuls do no ake ino accoun he acual performance of insurance socks, which ouperformed he markewide index during he sudy periods. From he heoreical perspecive, more imporanly, demuualized IPO socks should neiher ouperform nor underperform he benchmarks as he demuualizaion process alleviaes informaion asymmery and agency conflics in he IPO process. In he end, he resuls from his sudy show ha, once a proper benchmark is chosen, he posiive abnormal performance documened in he insurance lieraure disappears. In his sudy, he main focus is on he longrun performance of IPOs of a sricly defined sample of demuualized insurers. In paricular, only life and propery & casualy insurance firms are idenified from he oal sample. Oher ypes of insurance companies (e.g., medical malpracice insurance) are excluded from he sample. The jusificaion is ha hese companies do no face he same kind of compeiive environmen as life and propery & casualy insurance companies. Therefore, heir moivaion for demuualizaion and subsequen public offerings are no as clear as ha of life and propery & casualy insurance companies. More imporanly, only insurance firms ha conduc full demuualizaion are included in he sudy. 6 The reason is ha muual holding demuualizaion sill does no alleviae he agency conflics because original muual policyholders sill do no receive voing righs in he new sock companies. Unlike oher sudies which use markewide benchmark, more imporanly, his sudy examines he long-run performance of convering insurers in comparison wih a more direcly comparable se of benchmarks (insurance socks wih SIC Code and mached IPOs) in order o es he conjecure on he long-run underperformance of IPOs. In addiion, pos-ipo operaing performance, namely operaing cash flows, expense raios, and invesmen growh for he nex hree years are examined. By examining pos-issue operaing resuls, i can be concluded ha he need o access capial is wha moivaes demuualizaion and subsequen public offerings. V. Daa and Mehods To consruc he samples of life and propery & casualy IPO firms, a search was done on he Securiies Daa Corporaion (SDC) daabase for IPO firms wih SIC code Mos imporanly, seleced firms mus be posiively confirmed as life and propery & casualy insurance firms in Bes s Insurance Repors, Propery Casualy and Bes s Insurance Repors, Life Healh. A search was also done on he moneycenral.hoovers.com for IPO firms ha are described as life and propery & casualy firms for he period from July 1996 o December Abou wo-hirds of our firms are found via he 6 In a full demuualizaion, he policyholders surrender heir ownership righs while receiving compensaion in he forms of socks of he newly creaed company, cash or policy credis. In muual holding demuualizaion, a sock holding company, conrolled by a muual holding company, is creaed o direcly own a newly creaed sock insurance company. In his case, policyholders do no receive any shares of he new sock company. The conversion process for muual holding demuualizaion normally akes 6 o 12 monhs whereas he full demuualizaion could ake up o 2 years (Viswanahan and Cummins, 2003). 530

7 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 SDC daabase, while one-hird is found on he inerne. There are alogeher a oal of 46 insurance companies in he sample. The earlies IPO was conduced by Empire Fire and Marine Insurance in 1972, while he laes IPO was done by Specialy Underwriers Alliance Inc. in Among he sample firms, 17 firms are classified as operaing in life insurance business and 29 firms belong o he propery & casualy insurance business. The line of business is confirmed by Bes s Insurance Repors or by moneycenral.hoovers.com and business descripion for IPO firms from Securiies Daa Corporaion (SDC) daabase. Among he sample firms, 10 firms wen public immediaely afer demuualizaion according o Bes so heir conversion ype is classified as Y conversion. On he oher hand, 36 insurance firms are organized as sock firms since heir esablishmen so hey are classified as N conversion. The mos imporan aspec of sample selecion is ha he crieria are eiher verified by Bes or if here is no corresponding record of a conversion in Bes, he firm will be dropped from he sample. In he end, he sample of demuualized firms in his sudy is smaller han he sample of demuualized firms sudied in Lai, McNamara and Yu (2007) and Viswanahan (2006). Table 1 liss all he life and propery insurance firms, heir line of business, conversion ype and he IPO dae. Panel A of Table 2 repors he disribuion of IPOs over years and he disribuion of IPOs by line of business and conversion ype. From he yearly disribuion, a majoriy of he iniial public offerings occurred during he pas 15 years so i appears ha he period could be considered a ho marke years for he insurance indusry. [Inser Table 1 Here] [Inser Table 2 Here] Panel B of Table 2 presens he descripive saisics of IPO firms by line of business and ype of conversion. The issue size is calculaed by he offer price muliplied by he shares ousanding on he firs rading day. The median issue size for he overall IPO sample is $0.14 million. Life insurance IPO firms have higher median issue size of $0.78 million han ha of propery & casualy insurance IPO firms wih $0.08 million. Conversion IPO firms have median issue size of $1.14 million, while non-conversion IPO firms have smaller issue sizes wih a median $0.11 million. The median offer price for he whole IPO sample is $13.25 per share. The median offer prices are $17.00 per share and $12 per share, respecively, for life insurance and propery & casualy insurance IPOs. Conversion IPO firms have a median offer price of $15.63 per share while non-conversion IPO firms have a median offer price of $12.50 per share. Iniial reurn is measured as follows: Iniial Reurn = Firs Trading Day Price Offer Price Offer Price The overall median iniial reurn for insurance IPO firms is 5.36%. Life insurance IPO firms have a median iniial reurn of 5.00% and propery & casualy insurance IPO firms have a median iniial reurn of 5.73%. The median iniial reurn of conversion IPO firms is 7.12% while he median iniial reurn for non-conversion IPO firms is 4.84%. To evaluae he long-run performance of insurance IPO firms, he procedure in Rier (1991) is employed o calculae (1) he cumulaive average adjused reurns (CAR), where adjused reurns are compued using several differen benchmarks and (2) he 3-year buy and hold reurns for boh he IPO firm and a se of maching firms. For each IPO firm, he maching firm is seleced wih he same hree-digi indusry code as he IPO firm and wih he closes marke value a he end of he previous year of he IPO firm. Reurns are calculaed for wo inervals: he iniial reurn period (normally 1 day), defined as he offering dae o he firs closing price lised on he CRSP daily reurn files, and he afermarke period. The iniial reurn period is defined o be monh 0, and he afermarke period includes he following 36 monhs where monhs are defined as successive 21-rading-day periods relaive o he IPO dae. Thus, monh 1 consiss of even days 2-22, monh 2 consiss of even days 23-43, and so forh. For IPOs in which he iniial reurn period is greaer han 1 day, he monh 1 period is runcaed accordingly, e.g., if he iniial reurn period is 5 days, monh 1 consiss of even days For IPO firms ha are de-lised before heir 3-year anniversary, he afermarke period is runcaed. Monhly benchmark-adjused reurns are calculaed as he sock s monhly raw reurn minus he monhly benchmark reurn for he corresponding 21-rading-day period. The benchmarks used are (1) CRSP value-weighed NYSE-AMEX-NASDAQ index, (2) he CRSP value-weighed insurance indusry index (SIC code ), and (3) lised firms mached by indusry and size. The analysis based on he CRSP value-weighed NYSE-AMEX-NASDAQ index allows for a replicaion of resuls in previous insurance sudies (Lai, McNamara and Yu, 2007; Viswanahan, 2006). However, i is conended ha 531

8 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 he CRSP value-weighed insurance indusry index (SIC code ) is a more appropriae benchmark and provides a more conclusive comparison. To check for robusness, a group of financial firms mached by firm size is seleced as anoher benchmark. The benchmark-adjused reurn for sock i in even monh is defined as: AR = r r i i m The average benchmark-adjused reurn on a porfolio of n socks for even monh is he equally-weighed arihmeic average of he benchmark-adjused reurns: n AR 1 = n AR i i= 1 The cumulaive benchmark-adjused afermarke performance from even monh q o even monh s is he summaion of he average benchmarkadjused reurns: s CAR, q s = AR = q As an alernaive o cumulaive average benchmarkadjused reurns, 3-year holding period reurns are also calculaed: = 36 (1 + i r i = 1 ) R, where r i is he raw reurn on firm i in even monh. R i measures he oal reurn from a buy and hold sraegy where a sock is purchased a he firs closing marke price afer going public and held unil he earlier of (1) is 3-year anniversary, or (2) is delising. The 3-year oal reurn (Rier, 1991) and wealh relaives are defined as: 1+ average 3 year oal reurn on IPOs WR = 1+ average 3 year oal reurn on maching firms A wealh relaive of greaer han 1.00 can be inerpreed as an IPO firm ouperforming a porfolio of maching firms. In conras, a wealh relaive of less han 1.00 indicaes ha he IPO sock underperformed he benchmark. To evaluae pos issue operaing performance of insurance IPO firms, he examinaion of operaing performance measures is similar o ha in Jain and Kini (1994). There are hree measures of operaing performance used in his sudy. The firs measure is operaing reurn (EBIT) on asses, which is operaing income (before depreciaion and axes minus depreciaion) divided by oal asses a he end of he fiscal year (COMPUSTAT daa iem 13 divided by daa iem 6). The second operaing measure is operaing cash flow deflaed by oal asses a he end of he fiscal year. The raio equals income before exraordinary iems (COMPUSTAT daa iem 18) plus depreciaion (COMPUSTAT daa iem 14) divided by oal asses (COMPUSTAT daa iem 6). The change in operaing performance is calculaed as he median change in levels, i.e., he median value of {operaing reurn i () operaing reurn i (0)}, where i represens he firm, 0 represens he fiscal year prior of he IPO, and represens a pos-ipo fiscal year-end. The indusry-adjused change in operaing performance is calculaed by maching each IPO firm wih firms in he same indusry based on he hree-digi SIC code. Addiional operaing measure such as increase in invesmen (COMPUSTAT daa iem 113), and changes in expense raio (COMPUSTAT daa iem 189, which is he selling, general & adminisraive expenses, divided by daa iem 12, sales) are also employed in his sudy. Expense raio is a widely used measure of organizaional efficiency among insurance firms (Boose, 1990; Lai and Limpaphayom, 2003). Finally, several measures of invesor expecaions of pos-ipo earnings growh and he acual pos-issue earnings performance are examined. The resuls should show wheher invesors expec coninued earnings growh in he pos-issue period and if hese expecaions are fulfilled. Specifically, o sudy invesor expecaions of earnings poenial, he pos-issue M/B and P/E raios for boh insurance IPO firms and heir indusry counerpars are calculaed. The M/B raio of equiy is defined as he raio of marke value of equiy o he book value of equiy (COMPUSTAT daa iem 24 daa iem 25 divided by daa iem 60). The P/E is compued as COMPUSTAT daa iem 24 divided by daa iem 58. To measure he pos-issue earnings performance of insurance IPO firms and heir indusry counerpars, earnings per share (EPS) (COMPUSTAT daa iem 58 divided by daa iem 27) and pos-ipo changes are calculaed for he nex hree years. All he changes in hese raios are repored relaive o Year 0, he year of IPO. The nex secion presens and discusses he findings. VI. Resuls Table 3 presens he ARs and CARs of all of he insurance IPO firms in our sample. From Table 3, i can be seen ha he insurance IPO firms almos immediaely begin o underperform afer heir IPOs. The mean CAR becomes negaive in monh 3 and progressively becomes more negaive going from monh 1 o monh 36. The 3-year CAR of percen is saisically significan a convenional levels. Subsequenly, he insurance 532

9 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 IPOs are separaed ino subsamples by conversion classificaion (yes and no). Using he CRSP valueweighed NYSE-AMEX-NASDAQ index, Table 4 repors ARs and CARs of insurance firms ha wen public afer demuualizaion. Table 5 repors ARs and CARs for oher sock insurance firms ha also made iniial public offerings. From boh ables, i is apparen ha non-demuualized firms primarily drive he underperformance exhibied in Table 3. The resuls show ha demuualized IPO firms earn a large 56 percen 3-year reurn on average. Using he marke-wide benchmark (Table 4 and Table 5), he resuls are consisen wih he findings of Lai, McNamara and Yu (2007) and Viswanahan (2006) ha demuualized IPOs exhibi posiive abnormal reurns pos-issue. [Inser Table 3 Here] [Inser Table 4 Here] [Inser Table 5 Here] To check for robusness of previous sudies, addiional analyses are conduced using alernaive benchmarks o calculaed ARs and CARs. The alernaive benchmarks are he CRSP valueweighed insurance indusry index (SIC code ) and a randomly seleced group of lised financial firms mached by firm size. For each insurance IPO, 3 indusrial IPOs and 3 financial IPOs are also seleced, maching he sample firms wih same IPO year and closes firs rading day marke capializaion. Resuls for ARs and CARs calculaed using he CRSP value-weighed insurance indusry index (SIC code ) are shown in Table 6 and Table 7. [Inser Table 6 Here] [Inser Table 7 Here] The resuls wih differen bu appropriae benchmarks are quie sriking. The posiive longrun reurns of conversion IPOs drasically reduce when he benchmark is changed o he insurance index. For example, he long-run marke reurns in Table 6 are mildly posiive and no saisically significan when compared o secor reurns (SIC Code ). This resuls sill holds when using he mached financial firm reurns. This new finding indicaes ha he posiive performance documened by Lai, McNamara and Yu (2007) and Viswanahan (2006) migh have been oversaed by heir choice of he benchmark. In oher words, i appears ha he insurance secor as a whole was performing well when compared o he marke-wide measure. However, he abnormal reurns decline drasically when compared o firms in he same secor or o firms mached by financial characerisics. Table 7 also shows ha ARs and CARs for non-conversion insurance IPOs are negaive and saisically significan han when using he marke-wide benchmark. In he end, he noion ha he long-run marke performance of insurance IPOs ha jus wen hrough organizaion conversion is superior o he long-run marke performance of non-conversion insurance IPOs no longer holds. In conras, he empirical resuls provide suppor o he hypohesis ha conversion insurance IPOs do no exhibi superior or poor long-run marke performance when compared wih insurance socks or oher financial firms. The comparisons of resuls using differen benchmarks are graphically shown in Figure 1 and Figure 2. From Figure 1, he long-run performance of insurance IPOs was quie poor when compared o socks wih SIC Code and mached financial firms. Mos ineresingly, he insurance IPOs exhibi mosly posiive CARs when compared o IPOs from indusrial secors. Spliing he insurance IPO sample by conversion classificaion reveals a very ineresing finding. Figure 2 shows ha, compared o he NYSE-AMEX-NASDAQ Index, insurance firms ha jus wen hrough demuualizaion process immediaely before going public exhibi a superior long-run marke performance o oher ypes of insurance IPOs. This finding is consisen wih Lai, McNamara and Yu (2007) and Viswanahan (2006). However, he posiive abnormal reurns drasically reduce when he CRSP value-weighed insurance indusry index (SIC code ) is used as benchmark. The resuls using he mached-financial firms also reveal a similar paern. I is concluded ha demuualized insurance IPOs do no exhibi superior or poor performance when compare o insurance or mached-financial socks. Finally, Figure 3 shows ha demuualized life insurers exhibi superior performance when compared o non-conversion life insurers. This finding is consisen wih he noion ha life insurers benefi he mos from demuualizaion because hey are no longer consrain by he abiliy o raise funds o finance heir growhs. [Inser Figure 1 Here] [Inser Figure 2 Here] [Inser Figure 3 Here] Table 8 shows he pos-issue performance of IPO firms from a differen perspecive. Here, he cumulaive 3-year raw reurns of he insurance IPO firms are compared o he cumulaive 3-year raw reurns of a se of maching firms. The Wealh Relaive measure conrass he wo ses of reurns. From he resuls in Table 8, he performance on insurance IPOs is lower han ha of he maching firms bu he difference is no large. I is concluded ha here is no evidence of superior performance shown by demuualized insurance IPOs. Furher, i can be seen ha conversion IPOs ouperform non- 533

10 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 conversion IPO firms over he 3-year period (0.831 wealh relaive versus wealh relaive). [Inser Table 8 Here] Table 9 repors regression resuls. The main moivaion behind hese regressions is o see if conversion firms do indeed ouperform oher insurance firms, while conrolling for oher facors ha explain afermarke reurns. Following Rier (1991), he dependen variable is he 3-year oal raw reurn. Independen variables include he marke-adjused iniial reurns, marke reurns during he 3-year period, he size of he IPO proceeds, a dummy variable equal o 1 for insurance firms, a dummy variable equal o 1 for conversion firms, a dummy variable equal o 1 for non-conversion firms, and ineracion erms beween dummy variables. From Table 9, i is observed ha iniial reurns (IR) are negaively relaed o 3-year reurns. This finding is consisen wih ha of Rier (1991). If firms ime he marke when issuing heir IPOs, where firms issue IPOs when he marke overvalues hem, hen hese firms are likely o underperform in he long run. Iniially, he coefficiens for conversion insurance IPOs are posiive and saisically significan. However, he saisical significance of conversion dummy variable disappears when oher conrol variables are included in he models. I should be noed ha he coefficiens for non-conversion IPOs are negaive and saisically significan suggesing ha oher insurance IPOs underperform in he long-run. The finding ha conversion insurance firms do no underperform in he long run suggess ha hese demuualized insurers did no ake advanage of windows of opporuniy and ime he issues when share prices are overvalued. This resul provides suppor for he earlier conenion ha conversion firms have less laiude o ime he marke. In fac, conversion firms ouperform in he long run, suggesing ha hey go public for genuine economic-oriened reasons (i.e., o access capial for growh). The life insurance firm dummy is posiive and saisically significan, suggesing ha demuualized life insurance firms IPOs do especially well in he long run. This finding provides suppor o he noion ha life insurers benefi more from he demuualizaion as hey are able o obain sufficien capial o compee wih oher firms. [Inser Table 9 here] Operaing performance measures afer IPOs for all insurance companies are repored in Table 10. For all insurance IPOs, here is an iniial increase in operaing performance one year afer iniial public offerings (EBIT/TA increases), bu for he mos par operaing reurns evenually decline wo and hree years afer he IPO (boh EBIT and CF are negaive). These insurance firms also experience decreases in heir capial invesmens and increases in heir expense raios. Overall, he resuls show ha insurance firms, as a group, exhibi poor operaing performance afer iniial public offerings. Table 11 furher shows operaing performance by conversion ype. Panel A repors on conversion insurance firms and Panel B repors on non-conversion insurance firms. The differences in erms of pos-issue operaing performance are quie apparen. I appears ha non-conversion firms experience significan declines in operaing reurns and capial invesmens, and experience increases in expense raios. For conversion insurance firms, operaing performance and expense raios remain he same pos-ipo. These resuls provide imporan insighs as hey confirm he resuls for afermarke sock reurns. The capial raised by he IPO leads o sable operaing cash flows, invesmens, and expenses, which, in urn, lead o sable marke valuaions. Moreover, i appears ha managers of demuualized insurers are performing beer han he managers of oher ypes of insurers. The empirical evidence is consisen wih he hypohesis ha he agency conflic is relaively lower for conversion insurance companies which, in urn, leads o relaively beer pos-issue operaing performance. [Inser Table 10 here] [Inser Table 11 here] Finally, Table 12 shows marke expecaion measures. Panel A shows resuls for all insurance IPOs. Panels B and Panel C show resuls for conversion insurers and non-conversion insurers, respecively. The paerns are quie disinc. For non-conversion insurers, heir M/B raio, P/E raio, and EPS decline significanly pos-ipo. For conversion insurers, he M/B raio improves relaive o he indusry while P/E raios decline marginally. A he same ime, earnings per share measures for conversion insurance IPOs increase drasically. Overall, he resuls for marke expecaions sugges ha earnings growh of conversion insurers appear o coninue afer going public and ha expecaion by invesors is susained. The resuls also show ha expecaion of earnings growh for non-conversion insurance companies was no susained. [Inser Table 12 here] Overall, he empirical resuls show ha insurance companies ha wen public immediaely afer demuualizaion do no exhibi underperformance as observed among IPOs from oher indusries. More imporanly, he resuls of 534

11 Corporae Ownership & Conrol / Volume 8, Issue 2, Winer 2011, Coninued 5 his sudy are no consisen wih previous insurance sudies ha use marke-wide index as he benchmark. Once he appropriae benchmarks are used, he resuls show ha conversion insurance IPOs generae normal long-run reurns or perform as well as lised firms in he insurance secor or oher mached financial firms. The resuls also reveal ha insurance companies ha were already sock firms before public offerings exhibi poor long-run marke performance regardless of he benchmarks selecion. A possible explanaion for hese resuls is ha, due o he demuualizaion process, conversion insurers have lower informaion asymmery han oher ypes of firms ha make public offerings. Incidenally, hese conversion insurers also have relaively limied abiliy o ime he marke compared o oher ypes of IPOs. I is also possible ha he exen of agency conflics is lower for hese conversion insurance companies because he monioring mechanism improves afer conversion. Finally, he operaing performance measures of hese conversion insurers remain sable afer iniial public offerings. In he end, he empirical resuls provide indirec suppor o he noion pu forh by Rier (1991) ha firms ake advanage of windows of opporuniy by issuing shares when hey are overpriced and ha informaion asymmery leads invesors o be overly opimisic abou fuure prospec. By examining long-run performance of a group of companies ha wen public under an environmen wih relaively lower informaion asymmery and relaively lower agency conflics and documening no long-run underperformance among his group of IPOs, i is concluded ha wo major marke imperfecions, informaion asymmery and agency conflics, explain he underperformance of iniial public offerings documened by previous IPO sudies. VII. Conclusions The main objecive of his sudy is o examine he pos-ipo performance of insurance firms ha demuualized immediaely before going public. Demuualizaion is a process by which firms undergo a conversion from being a muual company (where policyholders own he firm) o becoming a sock company (where ouside invesors own he firm). The demuualizaion process is lenghy and ransparen, so hese conversion firms experience low informaion asymmery upon becoming a sock firm. In addiion, because ouside invesors are more acive owners han policyholders, hese conversion firms also experience lower agency coss upon becoming a publicly raded company. Consequenly, hese firms should experience a reducion in agency conflics beween managers and shareholders. Conversion firms usually cie he need o access capial for heir demuualizaion and subsequen public offering. Because he demuualizaion process is cosly wih respec o ime and effor, he benefis of becoming a sock company mus be significan. Given a low informaion asymmery and low agency cos environmen posdemuualizaion and pos-ipo, and given ha demuualizaion is a significan ransformaion, i is prediced ha demuualized IPO firms will no underperform in he long-run. A unique sample of demuualized and going-public insurance firms is carefully idenified in order o conduc he empirical ess. Previous insurance sudies have compared pos-ipo performance of demuualized insurance IPOs and find ha hese insurance IPOs exhibi superior performance when compared wih marke-wide benchmarks. In conras, his sudy documens ha demuualized firms do no ouperform a series of benchmark firm reurns, including insurance socks and oher financial firms. Furher, demuualized insurance firms also exhibi sable operaing reurns afer iniial public offerings. This laer finding is consisen wih he former finding. These resuls show, ex pos, ha demuualized firms are jusified in heir desire o demuualize and o subsequenly go public. Perhaps jus as imporan, he resuls also show indirec suppor for he window of opporuniy hypohesis. Because demuualized firms have less laiude o ime he marke, hey are less likely o be overvalued on he offering day. Regression resuls confirm he previous findings. Overall, he empirical findings suppor he hypohesis ha firms ha wen public in an environmen wih relaively low informaion asymmery and wih relaively low agency coss will no suffer from he same poor long-run underperformance ha has been documened in oher ypes of IPO firms. In summary, hese findings sugges, indirecly, ha informaion asymmery and agency coss explain he pos-issue underperformance ha has been documened in he finance lieraure. 535