1 Lemon Signaling in Cross-Listings Mihal Barzuza* This paper analyzes the deision to ross-list by managers and ontrolling shareholders assuming that they have private information with respet to the amount of private benefits of ontrol that they an extrat from their firms. Our model shows that sine ross-listing restrits the extration of private benefits of ontrol it serves as a signal that a manager or a ontrolling shareholder does not have many opportunities to extrat private benefits of ontrol. The analysis produes the following preditions. First, we find that while the opportunity to bond and signal low private benefits of ontrol enourages managers to ross-list, ounter intuitively, it may disourage ontrolling shareholders from rosslisting, sine suh a signal may derease the ontrol premium they an get for their shares. The model further predits that the value of the foreign firms in their home markets should derease but the value of the ontrol premium in those firms should inrease as a result of ross-listing by peers. Lastly, onsistent with the evidene, the model predits that there are no irumstanes in whih all firms ross-list, and that the firms that do not ross-list would be dominated by ontrolling shareholders. Our results have several impliations. First, they suggest that a weak bonding may result in U.S. apital markets attrating more firms than a strong bonding would, as was offered by a reent study and by an Interim Report of the Committee on Capital Markets Regulation. Seond, it predits that two markets may o-exist, one with a higher and one with a lower, inadequate level of dislosure. Thus, it may support some form of international regulation or harmonization. Third, our analysis suggests that a full onvergene around the world to effiient orporate governane paradigm is not likely to happen. Lastly and more generally, our results may suggest that the need in regulating ontrolling shareholders may be stronger than the need in regulating managers. Keywords: Cross-list, rivate benefits, Controlling shareholders, ontrol premium, Signaling *Assoiate rofessor, University of irginia Shool of Law. For their useful omments and suggestions I thank  and workshop and onferene partiipants at arvard, irginia, and the annual meetings of the Canadian, Israeli, and European law and eonomis assoiations.
2 . Introdution The last deade has witnessed the inreasing prevalene of ross-listing by foreign firms on U.S. exhanges. Typially, when ross-listing a foreign firm undertakes higher dislosure obligations and exposes itself to better enforement mehanisms than the ones it faed previously in its foreign market. Why firms ross-list has been the subjet of extensive literature. Equally interesting, and less disussed, is the question pertaining to why so many firms hoose not to ross-list, given proven benefits from ross-listing. The dominant hypothesis that explains ross-listing in the literature, the bonding hypothesis, asserts that managers and ontrolling shareholders ross-list in order to ommit to redue their extration of private benefits of ontrol (Coffee (999, 00, Stulz (999. That is, ross-listing is a way of onvergene to U.S. orporate governane standards. Although a signifiant body of evidene supports the fat that bonding is assoiated with ross-listing, there is evidene to suggest that it annot explain the ross-listing phenomenon in full. To start with, the firms that hoose to remain in their foreign markets exhibit a negative prie reation upon a ross-listing announement by one of their peers (Melvin and Tonone (004, Lee (003, whih seems to suggest that some form of signaling is taking plae. Seond, while ross-listing results in a signifiant positive prie reation the bonding itself is relatively weak (Liht (003, Siegel (00. Third, while many firms hoose to ross-list, many others hoose not to despite the signifiant positive prie reation, and more important, the deision to ross-list is inversely related to ontrol rights (Doidge et al (006. This paper develops a new explanation for ross-listing based on asymmetri information with respet to the extration of private benefits of ontrol. It suggests that by deiding whether or not to ross-list managers and ontrolling shareholders signal private information about the private benefits they an extrat from their firms. In partiular, it shows that when a manager or a ontrolling shareholder ross-lists he signals that he an not extrat a large amount of private benefits of ontrol from his firm. Our model shows that this signaling effet influenes managers and ontrolling shareholders inentives to ross-list in very different ways. Managers may benefit from signaling extration of small private benefits sine that will inrease the value of their shares. Controlling shareholders, ounter-intuitively, may benefit from onveying the opposite signal, that is, that they an extrat large private benefits of ontrol sine that would inrease the value of the ontrol premium they an get for their blok. Our analysis assumes that firms differ in how ostly it is to extrat private benefits from them. For instane, while in firms that operate in onentrated industries there may be lots of rents to extrat, in firms that operate in ompetitive industries extration of Other explanations inlude suggesting that firms ross-list in order to dismantle market segmentation and gain aess to more liquid markets (Foerster and Karolyi (999, and to inrease visibility to U.S. investors, (Lang et al., 003. Some proponents of the bonding hypothesis see this predition as supported by bonding (see Coffee (00. Sine bonding limits both managers and ontrolling shareholders from extrating private benefits of ontrol we think that the bonding hypothesis does not fully explain why ontrolling shareholders show signifiantly weaker tendeny to ross-list than managers. For a broader disussion of this issue see part 6 below.
3 private benefits may risk the ompetitiveness of the firm. Consequently, managers and ontrolling shareholders of the latter type of firms an extrat less private benefits of ontrol than managers and ontrolling shareholders of the former type. We also assume that these differenes in extration are not ompletely observable to the publi. The manager and the ontrolling shareholder have private information regarding how muh private benefits they an extrat and the fators that onstrain them from extrating private benefits of ontrol suh as the level of ompetition a speifi firm is faing. We onsider first a manager interested in selling some of his shares to the publi. Managers of all types, those that an extrat small and those that an extrat large private benefits of ontrol, would like to signal that they an extrat only small private benefits from their firm sine that would inrease the value of their shares. 3 Yet, managers who are able to extrat large private benefits of ontrol, have more to lose from migrating from a lax regime to a strit one, whih would ut some of their opportunities, than managers who do not have lots of opportunities to extrat private benefits of ontrol to begin with. Consequently, by tightening their restritions some managers redibly onvey information that they an extrat only a small amount of private benefits of ontrol from their firms. Modeling managers deision to ross-list we get the following results. First, the prie reation to ross-listing is larger under our hypothesis than under the bonding hypothesis, and the motivation for ross-listing is aordingly stronger. In addition, onsistent with the evidene and unlike the bonding hypothesis, our model predits that the value of firms that do not ross-list should derease as a result of ross-listing by peers. When we model the deision to ross-list by ontrolling shareholders we get markedly different results. Unlike managers, ontrolling shareholders an sell the opportunity to extrat private benefits of ontrol to others. Typially, in suh a transation the buyer pays a ontrol premium that reflets the opportunities to extrat private benefits from the firm (Dyk and Zingales (004, Barlay and olderness (989. As a result, signaling that a ontrolling shareholder is able to extrat large benefits from his firm would inrease the ontrol premium he an get and eventually may inrease the prie he an get for his ontrolling blok. A deision not to ross-list while other firms do would onvey the desirable signal. Analyzing the possible equilibria for ontrolling shareholders, for ases in whih the likelihood of selling the ontrol blok is signifiant 4 leads to the following results. First, our model shows that, in these irumstanes, an equilibrium in whih everyone rosslists does not exist. That is, it predits that there will always be ontrolling shareholders that hoose to remain in their home foreign markets. 5 These results are onsistent with the evidene that many firms do not ross-list and that the deision to ross-list is inversely related to ontrol rights (see Doidge et al (006. These results suggest that the 3 Suh a signal would inrease the value of the firms shares and in our model the manager holds some of his firm s shares. In reality, managers wealth is frequently tied to firm value to a ertain extent either beause they hold their firms shares, like in the model, or through bonuses or other rewards that are influened by performane. 4 The model assumes that the likelihood of selling the ontrol blok is suffiiently to make the prie that the ontrolling shareholder faes for his blok dereasing in. The preise ondition is ondition (8. 5 While under the bonding hypothesis it is possible that some firms do not ross-list it is also possible that everyone ross-lists.
4 U.S. apital markets annot hope to attrat all firms from foreign markets and that full onvergene to U.S. orporate governane standards is not likely to happen. Seond, in a separating equilibrium in whih some of the ontrolling shareholders ross-list our analysis shows that the value of the ontrol blok of the firms that remain in foreign markets may inrease as a result of ross-listing by peers, whih explains their deision to forgo the benefits of ross-listing. Why then would ontrolling shareholders ross-list at all? Controlling shareholders would ross-list only if by ross-listing they get benefits other than bonding, suh as greater liquidity and visibility to US investors. Thus, our model suggests that, when the likelihood of selling the ontrol blok is high, the bonding itself serves as a motivation not to ross-list for ontrolling shareholders. This is true even for ontrolling shareholders that extrat only small amount of private benefits of ontrol. In different than the bonding hypothesis we predit that if they do hoose to ross-list, it is despite of the bonding assoiated with ross-listing rather than beause of it. Reent studies have shown that the U.S. apital market is losing its ompetitive edge in attrating ross-listings (Zingales (006. Relying on this study an interim report released by the Committee on Capital Markets Regulation alled to ease regulation and enforement in the U.S. apital markets. 6 Yet, a debate is revolving over whether a derease in regulation and enforement is the right solution. 7 This artile suggests that indeed a strong bonding may deter some firms from ross-listing. Sine some of the firms ross-list despite of the bonding and beause of other benefits, they will ross-list only if the bonding is not partiularly strong. The above does not suggest that a weak bonding is desirable from a normative point of view but only that it ould help the U.S. to attrat more firms. In fat, our analysis shows that ontrolling shareholders may opt for suboptimal minority protetion. Thus, our analysis predits the emergene of two markets for dislosure, one with high level of dislosure and one with inadequate level of dislosure. These results an explain the emergene of the London Stok Exhange that offers lower dislosure standards as a main ompetitor to the U.S. apital markets and may support some form of international regulation or harmonization. In revealing a motivation not to ross-list our analysis has impliations for the debate over whether orporations around the world onverge to an effiient orporate governane paradigm. The analysis suggests that suh onvergene, either atual or funtional, is not likely to happen for all firms. Lastly, these effets should be taken into aount in disussions over the desirability of regulation in orporate law. Our analysis draws a sharp distintion between managers and ontrolling shareholders in suggesting that in the midstream stage of the firm s life managers have inentives to improve orporate governane while ontrolling shareholders have inentives to degrade orporate governane. Thus, it suggests that there may be more reasons to regulate the relationship between ontrolling shareholders and minority shareholders than the relationship between managers and dispersed shareholders. 6 See Interim Report of the Committee on Capital Markets Regulation (November 30, See igh and Low Finane: S.E.C. to Firms: Keep Money, Forget Rules Wall Street Journal, (De
5 Several studies have offered signaling based explanations for ross-listing. Fuerst (998 presents a model in whih firms move to striter regulatory regime to redibly onvey information on their firm s future prospets. Blass and Yafeh (00 similarly suggest that firms use osts assoiated with ross-listing to redibly onvey information about their value. Coffee (00 and Melvin and Tonone (004 suggest that by rosslisting firms signal high growth opportunities. Our analysis is different than theirs in that the signaling effet is reated beause of asymmetri information with respet to the extration of private benefits of ontrol rather than asymmetri information with respet to firm value or projets value or osts of ross-listing. Our results are markedly different as we predit opposite behavior by managers and ontrolling shareholders. Our analysis is also related to the literature on the way asymmetri information affets the adoption of orporate governane arrangements when firms first go publi. Bebhuk (00 has shown that in the presene of asymmetri information regarding firm value, owners might adopt rules with sub-optimal protetion to shareholders to signal the high value of their firm. Iaobui (00 showed that in the presene of suh asymmetries, firms might adopt exessive levels of investor protetion, also to signal high value. The urrent paper ontributes to this literature by showing how asymmetri information about private benefits, rather than firm value, affets the hoie of rules. Unlike previous literature, we show that whether asymmetri information leads firms to improve the effiieny of the legal rules that govern them or to worsen them depends on whether or not the firms has a ontrolling shareholder. The analysis proeeds as follows. art lays out the setting for the managers ase. art 3 analyzes managers deision to ross-list when information is symmetri. art 4 introdues asymmetri information to managers deision whether to ross-list. art 5 lays out the setting for the ontrolling shareholders ase. art 6 analyzes ontrolling shareholders deision whether to ross-list when information is symmetri. art 6 introdues asymmetri information to ontrolling shareholders deision whether to ross-list. art 7 disusses the empirial impliations of the analysis and offers a way to test it. art 8 derives impliations for the literature on ross-listing, the ompetitiveness of Amerian apital markets, the likelihood of onvergene, and the regulation of orporate law.. Framework of Analysis for Managers In this part we assume that there is no ontrolling shareholder, but only dispersed shareholders, that is, there is no shareholder who holds suffiiently large portion of the voting rights for whom it is profitable and feasible to effetively ontrol the firm. As a result, we assume that the manager makes the deision whether or not to ross-list. 8.. The Environment 8 Under U.S. state orporate law the board of diretors has the power to list or delist on and from exhanges (See Kahan (997. There is no reason to believe that the law is different in most of the ountries from whih firms ross-list on U.S. exhanges. Thus, when there is no ontrolling shareholder we assume that the manager is ontrolling the deision whether or not to ross-list. In parts 5-6 we assume that there is a ontrolling shareholder with suffiient ontrol to make the deision himself. 4
6 We onsider a four period setting. At T=0 the manager holds a fration of the firm ash flow, that is the model starts at the midstream stage of the firm life. There are two types of firms. From the first type of firms it is more ostly to extrat private benefits than from the seond type, and as a result managers of the first type would extrat less than managers of the seond type. The manager reeives a private signal regarding how ostly it is for him to extrat private benefits. The manager knows, based on the signal, how muh private benefits he an extrat without having the osts of extration for him outweigh its benefits. The uninformed investors do not know whether the manager an extrat high or low private benefits of ontrol but they hold a prior probability p that the manager an extrat only low private benefits of ontrol. At T= the manager sells a fration of his shares to the publi. rior to the sell he deides whether to ross-list or not. From his deision whether to ross-list or not investors draw inferenes on the amount of private benefits he an extrat. At T= the manager extrats private benefits of ontrol. At T=3 payoffs are realized to the manager and the investors. Our model builds on Bebhuk (00 whih shows that asymmetri information with respet to firm value may result in the hoie of suboptimal shareholder protetion at the IO stage. Like Bebhuk we assume that the extration of private benefits is ostly and that the manager and the ontrolling shareholder extrat the amount of private benefits that maximize their payoffs given the regime that governs their firm and the osts assoiated with the extration. Unlike Bebhuk we assume that some of the osts that are assoiated with the extration of private benefits of ontrol, and the extration itself, rather than firm value, are not observable to the publi. As a result, as the model shows managers and ontrolling shareholders may want to signal information about the private benefits of ontrol they extrat. This feature of the model is the essene of this paper T=0 The Manager olds a Fration of the Firm s Cash Flow. In the first period a foreign firm s shares are listed on its home market stok exhange. A manager of the foreign firm, who has an effetive ontrol of it, holds a fration of the firm s ash flow. The rest of the shares, namely a fration of - of the firm ash flow, are held by the publi. - <, that is, the model takes plae in the midstream stage of the firm s life. In this period the manager reeives a private signal with respet to what would be the osts of extrating private benefits from his firm, from whih he derives what is the amount of private benefits of ontrol that he an extrat from his firm. 0 The Investors do not have information on how muh private benefits the manager an extrat but they hold a prior probability p that the manager an extrat only low private benefits of ontrol.... T= The Manager Chooses a Legal Regime and Then Sells an Additional Fration of the Firm to the ubli 9 As this artile shows fousing on the extration of private benefits, rather than firms value, as an unobservable fator leads to qualitatively different results. An additional differene between our model and Bebhuk s model is that while he fouses on the IO stage we fous on the midstream stage. 0 For instane, a manager that reeives a signal that his firm is a type- firm know that the osts of extration for him are high and therefore he has less opportunities to extrat private benefits of ontrol than a manager of a type-l firm. 5
7 In the seond period the manager sells an additional fration of the firm to the publi. Like Bebhuk (00 we assume that prior to selling the manager has an opportunity to deide on the level of investor protetion that will govern his firm [, ]. We define to inlude the level of dislosure obligations and the level of enforement, and any other fators suh as analysts overage, that provide investors with protetion from extration of private benefits by managers. When selling additional shares to the publi the manager offers the ontrat (, in whih represents the prie he asks for the shares and represents the legal protetions desribed above. We assume ompetition in the market for investors so that the owner an get the prie he asks as long as this prie is not higher than the value of his shares to investors...3. T= The Manager Extrats rivate Benefits In the third period, after selling a fration to the publi, the manager extrats the amount of private benefits b that maximize the ex post value of his payments. The extration of private benefits is ineffiient, that is, it is assoiated with osts, so that if the osts to the firms of the extration are b the manager gets out of it only [ b L( b, C( b, ]. The osts that the extration of private benefits of ontrol imposes are of two types: ineffiieny osts assoiate with the extration that are independent of the legal regime, and osts that are assoiated with the legal regime that inhibits the extration. In assuming two kinds of osts we divert from existing models on extration of private benefits of ontrol whih assume that all of the osts from extration of private benefits are assoiated with the legal regime (see Bebhuk ( Ineffiieny Costs that are Independent of the Legal Regime Some of the ineffiieny osts that the extration of private benefits imposes are independent of the legal regime. For example, the extration of private benefits in a firm that operates in a ompetitive market inreases the likelihood that suh a firm will go bankrupt. Or, the opportunity to extrat private benefits may lead the manager to hoose a worse investment over a better one if the former allows him to extrat larger benefits than the latter. Or, if the manager extrats private benefits he may also have to distribute private benefits to others in the firm. As a result, even if the law does not limit the extration of private benefits of ontrol, we assume that suh extration imposes ineffiieny osts on the firm. These osts are denoted by C ( b,. We assume that these osts inrease with the extration of private benefits of ontrol. For the sake of simpliity we assume that C ( b, = b. C C C C This an be generalized as C ( 0, = 0, (0,, = 0, > 0 and that > 0. b b b We do not assume that the marginal osts of extration inrease with the level of private benefits of ontrol. To be sure, some of these marginal osts might inrease with the private benefits of ontrol. For instane, the higher the private benefits the manager takes the worse are the investments that he might undertake. 6
8 There are strong reasons to believe that firms differ in how diffiult it is to extrat private benefits of ontrol from them. Whereas in some ompanies it is easy to extrat large private benefits of ontrol in others it is almost impossible to do so without hurting the firm signifiantly (Roe (00. Following our risk of bankrupty example, while an extration of private benefits from firms in ompetitive industries ould impair the firm s likelihood to survive the ompetition, an extration of private benefits from firms with market power is likely to be less harmful and therefore is more likely to happen. Indeed, reent studies have shown negative orrelation between the extration of private benefit of ontrol and produt market ompetition (Guadalupe and erez-gonzalez (006. Similarly, while in some firms the manager an extrat private benefits primarily to himself without giving perks to others in others he may have to distribute some perks to others as well. Or in some firms the internal ontrol mehanisms are better than in others and therefore the manager does not extrat high private benefits of ontrol or otherwise would be fired. As a result the extration of private benefits is expeted to vary aross firms. The extration of private benefits is, for the most part, not observable, the very nature of suh extration is that it remains seret. Exposing the exat amount of private benefits of ontrol that the manager extrats and the ways in whih he extrats them may lead shareholders to blok the extration. Indeed empirial studies that assess private benefits of ontrol use indiret assessment by others, suh as ontrol premium that is paid for the sale of ontrol bloks (Dyk and Zingales (004, Barley and olderness (989, and the differenes in prie of shares with high and low voting rights (Nenova (003. The fators that lead to more or less extration are also, in part not observable. While investors generally know how ompetitive industries are, there is little information on what is exatly the level of ompetition a speifi firm is faing or to what extent the board is fulfilling its monitoring role. Thus we assume that the manager has private information on how muh he an extrat from his firm, and what are the osts of extration for him. In partiular, we assume C and b to be unobservable to the publi. To apture the unobservable differenes among firms the model inludes two types of ompanies. For the first type, whih is denoted by L, the extration of private benefits is relatively easy and therefore assoiated with relatively low osts (C L. For the seond type, whih is denoted by, the extration of private benefits is more diffiult and aordingly results in higher osts (C. The proportion of type L firms is p (0, and the proportion of type firms is thus -p. The type of the firm C is not observable to the market. In a separating equilibrium the market draws inferenes on a firm s type. In a Yet, in other ases the marginal osts might not inrease and even derease with the amount of private benefits of ontrol. For instane, the mere fat that the manager takes perks to himself might require him to distribute some perks to others. Yet, if he inreases his onsumption of private benefits he might not have to inrease it to others or in the worse ase will inrease it only in a linear way. Our qualitative results though do not hange signifiantly if we assume inreasing marginal osts. We also assume that ex post, the legal regime an trak, with some likelihood, partiular oasions of extration of private benefits of ontrol. Thus, if a manager extrats private benefits in an illegal way he may get aught. That is, with the investment of some osts a speifi extration is observable to the regulator ex post. 7
9 pooling equilibrium, where the market doesn t know the firm s type investors assume that a firm s osts are: C = pc ( p C L..3.. Costs Assoiated with the Legal Limitation on Extrating rivate Benefits of Control Seond, there are osts that are assoiated with and depend on the regulatory regime that the manager has adopted in the seond period, before selling his shares to the publi. Strit regulatory environment that requires high level of dislosure and involves high enforement inreases the risk that the ontrolling shareholder will be sued for extrating private benefits. Following Bebhuk (00 and to simplify the mathematial derivations, we assume that L( b, = b The Amount of rivate Benefits that the Manager Extrats Given the osts assoiated with the extration of private benefits of ontrol the manager extrats the amount of private benefits that maximizes the ex post value of his blok. As the first order ondition shows the private benefits that the manager extrats are dereasing in C. The intuition here is straight-forward. The higher the osts to the firm of extrating private benefits of ontrol, the less the manager (who holds some fration of the firm s ash flow will extrat suh benefits. 4 The manager s maximization problem is given by: Max b b b ( b b (. Solving for the F.O.C the level private benefits that the manager will extrat is b =. Sine the interest of this paper is in how private benefits affet the deision to rosslist we shall fous on the ases in whih the manager hooses to extrat some private benefits and therefore assume that ( > T=3 ayoffs are realized At T=3 ayoffs are realized for the manager and the investors. L L L L 3 This an be generalized as L ( 0, = 0, (0,, > 0, > 0 and > 0. As assumed b b b in Bebhuk (00 and Burkhart, anunzi, and Gromb (997, Not surprisingly, the private benefits that the manager extrats are also dereasing in. The higher the risk that the manager exposes himself to when he extrats private benefits of ontrol the less he tends to extrat private benefits. 8
10 ..4..ayoffs for the Manager Given that the manager extrats private benefits of ontrol in the size of ( b = the manager s ex post payoffs would be ( (3 π =..4..ayoffs for the Investors - The rie that Investors Would be Willing to ay for the shares The investors ex post payoffs, are equal to the value of the shares ex ante minus the harm aused by the private benefits that the manager is antiipated to extrat: (4 π s = I =. Sine investors antiipate the extration of private benefits of ontrol this is also the prie I, that they would be willing to pay for the fration of shares. That is they would disount the prie they are willing to pay for the shares to reflet the inreased private benefits that the manager will extrat. Note that the payoffs to the investors, and as a result the prie that they will pay for π s the fration of shares, are inreasing in : = > 0. Consequently, the investors would be willing to pay more for shares of firms that ross-list. The payoffs to investors are also inreasing in the firm s type. Reall that the firm type desribes the osts assoiated with the extration of private benefits. In a Type- firm, for instane, the osts assoiated with the extration of private benefits are high and as a result the manager is likely to extrat less private benefits of ontrol. Investors are willing to pay more for this firm (whose manager takes less to his poket, than for a Type-L firm in whih the manager is expeted to extrat higher private benefits of ontrol. 3. Managers - Symmetri Information The Bonding ypothesis In order to demonstrate the signaling effets we shall fous first on the symmetri information ase as a benhmark. We analyze what would happen if the publi knew for eah firm the osts of extrating private benefits, and therefore the amount of private benefits its manager extrats. The symmetri information model demonstrates that, as was offered by Coffee (999, 00, some managers hoose to ross-list in order to bond themselves to a striter legal regime and ommit to extrat small private benefits of ontrol. It also shows, as offered by Coffee (999, 00 that some managers may hoose not to ross-list. 9
11 In hoosing the legal regime the manager, knowing that investors antiipate his ex post behavior, should pik the regime that maximizes his ex ante payoffs, whih payoffs onsist of the private benefits he extrats, and the value of his shares: ( (5 π = b b b ( ( b I =. The FOC that determines the legal regime that the manager will hoose is: π (6 = As the FOC shows the profitability of ross-listing to the manager depends on the level of the ineffiieny osts that result from his extration of private benefits. We an derive the minimum osts that are neessary for a manager to fae in order to ross-list: ROOSITION : Under symmetri information, managers that ross-list are those whose is large enough to satisfy the following ondition: (7 > The intuition for this result whih is proved in the appendix is the following. The higher osts of extration for a manager from his firm, the less private benefits the manager expets to extrat, and in turn the less he loses from migrating to a striter regime that limits his extration. For suffiiently high osts the inrease in the market prie of his shares would outweigh the loss of private benefits to the manager and motivate him to ross-list. Thus, we show that when there are differene is osts of extration, onsistent with the preditions of the bonding hypothesis (see Coffee (999, 00 managers that would ross-list are those that have higher osts and therefore fewer opportunities to extrat private benefits of ontrol. In the following parts we demonstrate that introduing asymmetri information to the model strengthens the motivation to ross-list for managers. Our analysis of asymmetri information builds on the bonding hypothesis in assuming that by ross-listing managers and ontrolling shareholders ommit to extrat less private benefits of ontrol. We are analyzing how the introdution of asymmetri information affets firms motivation to bond. 4. Managers - Asymmetri Information the Signaling of rivate Benefits ypothesis Under asymmetri information investors do not observe the type of the firm, that is, they do not observe how ostly it is to extrat private benefits from eah firm and how muh eah manager extrats as a result. 5 This part analyzes two possible equilibria that 5 The onept of equilibrium that we adopt is a erfet Bayesian Equilibrium with a refinement of the Cho-Kreps intuitive riterion. A set of strategies and a belief funtion onstitute a BE iff: 0
12 may result. 6 A separating equilibrium in whih some managers ross-list and some do not ross-list or a pooling equilibrium in whih all managers ross-list. 7 Analyzing the properties of a separating equilibrium we find that the prie reation to ross-listing under asymmetri information (the signaling of private benefits hypothesis is stronger than the prie reation to ross-listing under symmetri information ( the bonding hypothesis. Seond, and related, we show that the threshold for ross-listing is lower under asymmetri information. That is, firms that would not ross-list under symmetri information might do so under asymmetri information. Lastly, we show that for firms that remain in their home markets the value of the shares should derease as a result of ross-listing by peers. 4.. There is no ooling Equilibrium in whih everyone Remains in Foreign Markets ROOSITION : There is no pooling equilibrium in whih everyone stays in foreign markets Our first step is to show that there is no pooling equilibrium in whih all of the managers hoose to remain in foreign markets. The proof of this result is available in the appendix. In fat as the proof shows the only possible pooling equilibrium is one in whih all managers ross-list on the stringent regime, the one with the highest. 4.. A Separating Equilibrium Managers Signal Low rivate Benefits of Control One possible equilibrium is a separating equilibrium in whih some managers rosslist and others do not. A separating equilibrium will result if and only if managers of Type-L ompanies hoose not to mimi the managers of Type- ompanies. This leads to the following ondition. ROOSITION 3: if managers of Type-L ompanies prefer their ineffiient symmetri ontrat ( (, on the ontrat ( (,, 8 then the unique solution is a separating L (a The manager s strategy is optimal for him given the investors strategy. (b Given their beliefs about whih type of manager they fae the investors hoie whether to aept or rejet a ontrat is optimal for them. ( The belief funtion is derived from the managers strategy using Bayes law when possible. (d Investors beliefs satisfy the Cho-Kreps intuitive riterion. An equilibrium satisfies the intuitive riterion if the bad type of manager would not adopt an out of equilibrium ontrat if he does worse with this ontrat, relative to its expeted equilibrium utility even if that would lead him to be onsidered as a good type and the good type manager would adopt suh a ontrat if he would do better with this ontrat, relative to its expeted equilibrium utility if adopting this ontrat would lead him to be reognized as the good type. The Cho-Kreps riterion is aepted as a refinement of BE sine when only one type an possibly benefit from deviation, then onstrution of beliefs that the other side has deviated upon witnessing a deviation is not reasonable. 6 There is an additional possible equilibrium that we do not disuss here, a hybrid equilibrium. 7 There is another possible equilibrium a hybrid equilibrium whih we do not analyze here. 8 This ondition ensures that the manager of a Type-L firm has no inentive to mimi the manager of a Type- firm. For a separating equilibrium to hold the ondition doesn t have to inlude also a requirement that the manager of a Type- firm is better off not pooling with the manages of Type-L firms sine as shown in proposition the only pooling equilibrium that an exist is an equilibrium on the stritest legal
13 equilibrium in whih managers of Type-L firms are offering their ineffiient symmetri information ontrat ( L (, and the managers of Type- firms are offering a ~ ~ ~ [( L ( ] ontrat (, ( suh that = ( ( The proof for this result is available in the appendix. The following is an analysis of the properties of a separating equilibrium rie Reation to Cross-Listing is Stronger Under the Asymmetri Information Case Under the bonding hypothesis prie reation to ross-listing should reflet the effiieny gains assoiated with it. Introduing asymmetri information we get the following result: Corollary : In a separating equilibrium prie reation to ross-listing is greater than the effiieny gains from ross-listing and the expeted prie reation under the bonding hypothesis. 9 This result, whih is proved in the Appendix, is onsistent with the evidene as explained in part 7 below. The intuition for this result is as follows. Under the bonding hypothesis the market prie reation to ross-listing reflets the derease in the private benefits a manager an extrat as a result of ross-listing and the ineffiieny osts they impose. Under the signaling of private benefits hypothesis the prie reation reflets these effets plus an additional effet. A manager s deision to ross-list reveals that he in general extrats less private benefits of ontrol. The prie reation therefore also reflets the information that he will extrat a small amount of private benefits under the new regime The Threshold for Cross-listing is Lower under the Asymmetri Information Case The bonding hypothesis predits that managers will ross-list only if their share in the savings to the firm from ross-listing outweighs the redution in their private benefits of ontrol. Under the signaling of private benefits hypothesis, however, managers might hoose to ross-list even in ases where their private benefits are higher than their share in the savings to the firm from ross-listing. In partiular, as the next orollary suggests: Corollary : Firms that would not ross-list under symmetri information might rosslist under asymmetri information in order to signal their type. L L regime. Thus, if the managers of type-l ompanies hoose not to ross-list there will be a separating equilibrium. 9 Notie that we did not assume that ross-listing is assoiated with osts. If it is then orollary should only say that the prie reation to ross-listing is higher in the asymmetri information ase than in the bonding ase.
14 The intuition for this result, whih is proved in the Appendix, is as follows. By rosslisting managers get additional gains from revealing their type, that is, revealing that they in general extrat relatively low private benefits of ontrol, sine it inreases the value of their shares. These gains help to offset the loss of some private benefits to the manager. C, is lower Thus, the threshold ost for ross-listing under asymmetri information ( CL A than the threshold ost for ross-listing under symmetri information ( C. CL S CCL A C CL S C 4.5. Share alue of Firms that Do Not Cross-List Should Fall Another result of proposition relates to firms that do not ross-list. As the market learns information on the managers that ross-list it also learns information on the managers that do not to ross-list. In partiular, the market learns that managers that remain in their home markets extrat more private benefits than their peers in similar firms that have ross-listed. Corollary 3: Share value of firms that do not ross-list should fall. This result, whih is proved in the appendix is onsistent with and explain the evidene as explained in part 7 below There is a ossible ooling Equilibrium in Whih All Managers Cross-List A separating equilibrium is not the only possible equilibrium. It is also possible that all managers would ross-list, that is, a pooling equilibrium on the effiient regime. This would happen if the advantages of being onsidered as a good type outweigh, for the bad type manager, the osts of ross-listing on the stritest legal regime. In this ase the only possible equilibrium is a pooling equilibrium on the stritest legal regime. ROOSITION 4: If managers of Type-L firms prefer the ontrat ( (, on their symmetri information ontrat ( L (, then the unique solution onsists of both types hoosing to ross-list and offer the ontrat ( (,. The proof for this result is available in the appendix roperties of Separating and ooling Equilibria for Managers 3
15 The table below summarizes the properties of the possible separating and pooling equilibria for managers. ross- Who lists Who does not ross-list Results Separating equilibrium Managers with lower private benefits Managers with higher private benefits - igher prie reation to rosslisting than under bonding hypothesis - Threshold for ross-listing is lower than under bonding hypothesis - Negative prie reation to not ross-listing ooling equilibrium in whih everyone ross-lists All of the managers No one Everyone rosslists on the stringent regime No other pooling equilibrium At least some managers rosslist No managers, or some managers but not everyone No equilibrium in whih everyone does not ross-list 5. Framework of Analysis - Controlling Shareholders In this part we assume that the firm has a ontrolling shareholder that holds suffiiently large portion of shares that allows him to have ontrol over his firm and therefore he makes the deision as to whether or not to ross-list. Unlike managers, ontrolling shareholders an sell the opportunity to extrat private benefits of ontrol by selling their ontrol blok. 0 This part introdues this possibility. 5.. The Environment The setting is very similar to the manager s setting we disussed in setion exept for one thing: After he hooses the legal regime and sells a fration of his shares a 0 To be sure, managers sometimes get golden parahutes for losing their ontrol, whih may be viewed as ontrol premium. Yet, it is not lear that these golden parahutes reflet the private benefits that managers extrat from their firms sine managers many times have influene on setting their own golden parahutes. Moreover, even if they did there are less reasons to suggest that managers need to signal this kind information to their own board. 4
16 liquidity shok that might our with probability q leads the ontrolling shareholder to sell his ontrol blok. In transferring the blok the ontrolling shareholder also transfers the power to extrat private benefits from the firm. B represents the prie a ontrolling shareholder will reeive for his ontrolling blok and I represents the prie that he will get for the publily traded shares he sells to investors. For the sake of simpliity and without loss of generality it is assumed that the ontrolling shareholder has all of the bargaining power when selling his blok. Given the possibility of suh a liquidity shok the ontrolling shareholder s ex ante payoff is: ( (6 π = I qb ( q I B = ( b = b b b ( ( b In hoosing a legal regime the ontrolling shareholder takes into aount both the prie that he will get for the shares he sells to the publi and the prie that he will get for his blok. While I, the prie that investors will pay for the publi shares, is inreasing in, B, the prie that a potential buyer will pay for the ontrol blok, is dereasing in. = I q B = ( q To demonstrate the differene between the ase of a ontrolling shareholder and the ase of a manager we will fous on the ases in whih the likelihood of selling his blok, q, is suffiiently high for the ontrolling shareholder so that the effets of selling the ontrol blok with high premium outweigh the effets of selling the fration to the publi in high prie, in the sense that the overall prie that he is seeing is dereasing in : Sine = ( q (. If (8 < q for every, then the prie for both types of ontrolling shareholders is dereasing in. Thus for the rest of the analysis we assume that ondition 8 is met. 5
17 6. Controlling Shareholders Symmetri Information The symmetri information ase for the ontrolling shareholder is idential to the symmetri information ase for the manager. Thus, as this part shows the ondition for ross-listing under the symmetri information ase is the same as ondition (, the ondition for ross-listing for managers: ROOSITION 5: Under symmetri information, ontrolling shareholders will ross-list if their is suffiiently large to satisfy: (7 > This result suggests that there is no differene under the bonding hypothesis between managers and ontrolling shareholders. This result stands in ontrast to the literature that uses the bonding hypothesis to explain the observation that ontrolling shareholders have less motivation to ross-list than managers. John Coffee, who was the first to raise the bonding hypothesis, suggested that ontrolling shareholders may have less motivation to ross-list than managers sine ontrolling shareholders want to extrat private benefits of ontrol, while firms with no ontrolling shareholders tend to maximize shareholder value (Coffee (00. Yet, managers also want to extrat private benefits of ontrol and sine the board typially has the power to make listing deisions (see Kahan (997 the ageny problem here is as apparent as in the ontrolling shareholder ase. Coffee also argues that ontrolling shareholders will have less motivation to ross list sine they get ontrol premium for their blok. As long as we assume symmetri information however, as the bonding hypothesis does, there is little differene between shareholders and managers also in that respet. The osts for a manager and a ontrolling shareholder of limiting the future extration of private benefits of ontrol is the present value of the private benefits of ontrol he is expeted to extrat, whih is exatly the ontrol premium a potential buyer would be willing to pay. Thus, under the bonding hypothesis there is almost no differene between a ase in whih there is a sale of ontrol blok and a ase in whih there isn t, and aordingly no differene between managers and ontrolling shareholders. The different arises only if the ontrol premium a potential buyer would pay to the ontrolling shareholder depends on the information he has with respet to the private benefits he an extrat as a ontrolling shareholder in the firm. In that ase the ontrolling shareholder may hoose not to ross-list in order to signal that he extrats high private benefits of ontrol. As the following part demonstrates the asymmetri information assumption explains why ontrolling shareholders ross-list less than managers. This is true sine the symmetri information ase for the ontrolling shareholder is idential to the ase in whih the ontrolling shareholder keeps the blok and extrats the private benefits with ertainty. To be sure, there may be a differene if the manager position is expeted to be terminated in the near future. In that ase the horizon that he sees for extrating private benefits of ontrol is limited, yet, as long as this is not likely to happen in the near future disounting for time will make the differene between a manager and a ontrolling shareholder almost negligible. 6
18 7. Controlling Shareholders Asymmetri Information Signaling a Lemon This part analyzes the ontrolling shareholder hoie whether to ross-list under asymmetri information. 3 In onverse to the ase of a manager (namely, a firm with no ontrolling shareholder and ounter-intuitively, a ontrolling shareholder may want to signal that he extrats high private benefits of ontrol. When the market realizes that a ontrolling shareholder is of a bad type, that is, one that extrats high private benefits of ontrol, potential buyers may be willing to pay a higher ontrol premium for his ontrol blok. As a result, when the likelihood of selling the blok is suffiiently high we get the following results: First, there are no equilibria in whih everyone ross-lists. There will always be ontrolling shareholders who would stay in their home market rather than ross-list. This result is onsistent with and explains the evidene as disussed in part 8 below. Seond, a separating equilibrium in whih the ontrolling shareholders that extrat less private benefits ross-list and those that extrat more private benefits remain in their home markets may emerge. In suh equilibrium the value of the ontrol premium, as well as the value of the ontrol blok, are inreasing upon ross-listing by peers. As a result, suh equilibrium may emerge only if there are some advantages to ross-listing other than bonding. ut differently, ontrolling shareholders will not ross-list just in order to bond. Lastly, a pooling equilibrium in whih none of the ontrolling shareholders rosslists may emerge. That is, from some ountries we may view no ross-listings by ontrolling shareholders but only by managers. 6.. There is no Equilibrium in whih everyone Cross-Lists ROOSITION 6: Under ondition 8 the only pooling equilibrium is one in whih all firms do not ross-list. The intuition for this result whih is proved in the appendix is shown graphially: ( ( The pooling line is = q. Note that from (7 the pooling line is dereasing in. The ontrolling shareholders iso-rofit urves are: ( = π ( q ( Note that = q that is, hat the Type-L urve is steeper than the urve of type-. 3 We assume that when there is a ontrolling shareholder he ontrols the manager so that he makes the deision whether or not to ross-list. 7
19 For every point on the pooling line other than = there is another point with a lower whih is better for Type-L than remaining on the pooling line and worse for Type- than remaining on the pooling line. As a result, any equilibrium on the pooling line other than = is not stable. L ( ( C C L p ( 8
20 6.. Controlling Shareholders will Cross-List only if there are Advantages Other than Cross-Listing ROOSITION 7: For q that is suffiiently large to meet ondition (8 a separating equilibrium will exist only if ross-listing is assoiated with advantages other than bonding. The intuition for this result whih proof is omitted is the following: Sine the prie is dereasing in then both ontrolling shareholders prefer to be in the foreign market. The signaling effet only adds to that sine both ontrolling shareholders would benefit from signaling high private benefits of ontrol. Following are the properties of suh a separating equilibrium Control blok alue of Firms that Remain in Their ome Market Should Rise Corollary 4: The ontrol blok value of firms that remain in their home market should rise The intuition for this result whih is proved in the appendix is the following: When firms ross-list potential buyers of the ontrol blok realize that the ontrolling shareholders of those firms extrat a relatively small amount of private benefits of ontrol. As a result the potential buyers would be willing to pay only a low ontrol premium that reflets this information. otential buyers also realize that firms that remain in home markets and forgo the benefits of ross-listing do so beause they extrat a high amount of private benefits of ontrol. As a result, potential buyers would be willing to pay more for ontrol bloks of firms that remain in their home markets as their peers ross-list The threshold for Cross-Listing is igher under the Symmetri Information Case Corollary 5: Firms that would ross-list under symmetri information might not do so under asymmetri information. The intuition for this result, whih is proved in the Appendix, is as follows. By rosslisting ontrolling shareholders might suffer from revealing their type, that is, revealing that they in general extrat relatively small private benefits of ontrol. While suh information inreases the prie investors should be willing to pay it dereases the ontrol premium potential buyers would pay for the blok. If the likelihood of selling the ontrol blok is suffiiently high a ontrolling shareholder would prefer to signal that he extrats a large amount of private benefits of ontrol by staying in his home market. 4 Note that it is not only that the value of the premium is inreasing but also the value of the blok is inreasing. The reason is that every ontrolling shareholder will extrat private benefits only to the point that maximize the value of his blok, that is no ontrolling shareholder will extrat private benefits if the harm that is aused to his shares by the extration is larger than the private benefits he gets. 9