Globalizing Corporate Governance: Convergence of Form or Function

NELLCO NELLCO Legal Scholarship Repository Harvard Law School John M. Olin Center for Law, Economics and Business Discussion Paper Series Harvard Law School 10-7-1999 A Theory of Path Dependence in Corporate Ownership and Governance Lucian Bebchuk Harvard Law School Mark Roe Harvard Law School Recommended Citation Bebchuk, Lucian and Roe, Mark, "A Theory of Path Dependence in Corporate Ownership and Governance" (1999). Harvard Law School John M. Olin Center for Law, Economics and Business Discussion Paper Series. Paper 266. http://lsr.nellco.org/harvard_olin/266 This Article is brought to you for free and open access by the Harvard Law School at NELLCO Legal Scholarship Repository. It has been accepted for inclusion in Harvard Law School John M. Olin Center for Law, Economics and Business Discussion Paper Series by an authorized administrator of NELLCO Legal Scholarship Repository. For more information, please contact [email protected]. ISSN 1045-6333 A THEORY OF PATH DEPENDENCE IN CORPORATE OWNERSHIP AND GOVERNANCE Lucian Arye Bebchuk Mark J. Roe Discussion Paper No. 266 10/99 Harvard Law School Cambridge, MA 02138 The Center for Law, Economics, and Business is supported by a grant from the John M. Olin Foundation. www.law.harvard.edu/Programs/olin_center forthcoming in Stanford Law Review, November 1999 A Theory of Path Dependence in Corporate Ownership and Governance Lucian Arye Bebchuk* and Mark J. Roe** Corporate structures differ among the advanced economies of the world. We contribute to an understanding of these differences by developing a theory of the path dependence of corporate structure. The corporate structures that an economy has at any point in time depend in part on those that it had at earlier times. Two sources of path dependence—structure driven and rule driven—are identified and analyzed. First, the corporate structures of an economy depend on the structures with which the economy started. Initial ownership structures have such an effect because they affect the identity of the structure that would be efficient for any given company and because they can give some parties both incentives and power to impede changes in them. Second, corporate rules, which affect ownership structures, will themselves depend on the corporate structures with which the economy started. Initial ownership structures can affect both the identity of the rules that would be efficient and the interest group politics that can determine which rules would actually be chosen. Our theory of path dependence sheds light on why the advanced economies, despite pressures to converge, vary in their ownership structures. It also provides a basis for why some important differences might persist. Key words: corporate ownership; corporate governance; path dependence; corporate law; comparative law; JEL classification: G3, K22. *. William J. Friedman and Alicia Townsend Friedman Professor of Law, Economics, and Finance, Harvard Law School. ** Milton Handler Professor of Business Regulation, Columbia Law School. We benefited from the comments of Merritt Fox, Ron Harris, Marcel Kahan, Ehud Kamar, Louis Kaplow, Randy Kroszner, Benjamin Mojuye, Roberta Romano, Reinhard Schmidt, Gerald Spindler, Luigi Zingales, and workshop participants at Harvard, the University of Frankfurt, New York University, the 1997 Columbia Law School conference on convergence of systems, the 1998 meetings of the American Law and Economics Association and the Summer Institute of the National Bureau of Economic Research, and the 1999 European Symposium on Financial Markets in Gerzensee. For financial support, Lucian Bebchuk thanks the John M. Olin Center for Law, Economics, and Business at Harvard Law School and the National Science Foundation, and Mark Roe thanks the Columbia Law School Sloan Project on Corporate Governance. TABLE OF CONTENTS INTRODUCTION ..................................................................................................................... 1 I. Explaining Persistent Differences ................................................................................. 4 A. The Focus of Our Inquiry........................................................................................ 4 B. The Persistence of Corporate Differences .............................................................. 6 1. Globalization and the drive toward efficient structures.................................... 6 2. Persistence ........................................................................................................ 7 C. Sources of Path Dependence................................................................................... 8 II. Structure-Driven Path Dependence............................................................................. 10 A. Path Dependence of the Efficient Structure .......................................................... 10 1. Sunk adaptive costs ......................................................................................... 10 2. Complementarities .......................................................................................... 11 3. Network externalities....................................................................................... 11 4. Endowment effects........................................................................................... 12 5. Multiple optima ............................................................................................... 12 B. Persistence of Existing Structures Due to Rent-Seeking....................................... 13 1. Persistence of concentrated ownership........................................................... 13 2. Persistence of diffuse ownership ................................................................. …19 3. Persistence of German codetermination ......................................................... 20 4. Persistence in the face of globalization........................................................... 20 C. Conclusion on Structure-Driven Path Dependence .............................................. 22 III. Rule-Driven Path Dependence .................................................................................... 23 A. Systems of Corporate Rules................................................................................... 23 B. Path Dependence of the Efficient Rules ................................................................ 24 1. Sunk costs and complementarities................................................................... 24 2. Multiple optima ............................................................................................... 25 C. Path Dependence of the Rules that Are Actually Chosen ..................................... 25 1. Initial conditions and the political economy of corporate rules ..................... 26 2. Rules affecting concentrated and diffuse ownership structures...................... 26 3. Globalization and the pressure to adopt efficient rules .................................. 29 4. Can contracts generally substitute for legal rules?. ....................................... 29 5. Reincorporations............................................................................................. 30 6. Public-regarding victories over interest group politics.................................. 31 D. Elimination of Differences in Rules by Political Fiat ........................................... 32 E. Conclusion on Rule-Driven Path Dependence...................................................... 34 IV. Other Bases for Persistent Divergence........................................................................ 34 A. Differences of Opinion .......................................................................................... 34 B. Differences in Firms and Markets......................................................................... 35 C. Differences in Culture, Ideology, and Politics..................................................... .36 Conclusion............................................................................................................ ...…..…37 A Theory of Path Dependence in Corporate Ownership and Governance Lucian Arye Bebchuk and Mark J. Roe © 1999 Lucian Arye Bebchuk and Mark J. Roe. All rights reserved. INTRODUCTION Corporate ownership and governance differ among the world’s advanced economies. Some countries’ corporations are diffusely owned with managers firmly in control, other countries’ corporations have concentrated ownership, and in still others, labor strongly influences the firm. During the past half-century since World War II, economies, business practices, and living standards have converged in Western Europe, the United States, and Japan. But their corporate ownership structures have remained different, and different degrees of ownership concentration and labor influence have persisted. What explains these differences? And should they be expected to persist or to disappear? We shed light on the above questions by showing that there are significant sources of path dependence in a country’s patterns of corporate ownership structure. Because of this path dependence, a country’s pattern of ownership structures at any point in time depends partly on the patterns it had earlier. Consequently, when countries had different ownership structures at earlier points in time—because of their different circumstances at the time, or even because of historical accidents—these differences might persist at later points in time even if their economies have otherwise become quite similar. In Part I, we describe our inquiry. Why, against the background of the forces for global convergence, do the advanced economies differ so much in their corporate ownership structures? For concreteness, our analysis focuses on one important dimension of differences among countries: whether their corporations commonly do or do not have a controlling shareholder. We distinguish in Part I between two sources of path dependence. One source of path dependence—which we label structure-driven path dependence—concerns the direct effect of initial ownership structures on subsequent ownership structures. We show how the corporate structures that an economy has at a given point in time are influenced by the corporate structures it had earlier. Another source of path dependence—which we label rule-driven path dependence—arises from the effect that initial ownership structures have on subsequent structures through their effect on the legal rules governing corporations. By corporate rules, we mean all the legal rules that govern the relationship between the corporation and its investors, stakeholders, and managers and the relationships among these players—including not only corporate law as conventionally defined but also securities law and the relevant parts of the law governing insolvency, labor relations, and financial institutions. Corporate rules themselves, we show, are path dependent. The fol1 lowing two Parts of the paper analyze in turn these two main sources of path dependence. Part II focuses on structure-driven path dependence. Here we analyze how choices of corporate ownership structure will be directly influenced by the initial 1 ownership structures that the economy had. To this end, we show how choices of ownership structure might differ in two economies that now have identical corporate rules but started with different ownership structures. We identify two reasons why prior ownership structures in an economy might affect subsequent structures—one grounded in efficiency and the other in rent-seeking. First, the efficient ownership structure for a company is often path dependent. Due to sunk adaptive costs, network externalities, complementarities, and multiple optima, the relative efficiency of alternative ownership structures depends partly on the structures with which the company and/or other companies in its environment started. Second, existing corporate structures might well have persistence power due to internal rent-seeking, even if they cease to be efficient. Those parties who participate in corporate control under an existing structure might have the incentive and power to impede changes that would reduce their private benefits of control even if the change would be efficient. For example, a controlling shareholder might elect not to move her firm to a diffused ownership structure because the move would reduce the controller’s private benefits of control. Similarly, the managers of a company with diffused ownership, seeking to maintain their independence, might elect to prevent their firm from moving to a concentrated ownership structure even if the move would be efficient overall. And in nations in which labor unions play a role in corporate control, union leaders might seek to maintain structures that give them such power. As long as those who can block structural transformation do not bear the full costs of persistence, or do not capture the full benefits of an efficient move, inefficient structures that are already in place might persist. To be sure, all potentially efficient changes would take place in a purely Coasian world. However, as we show, the transactions feasible in our imperfectly Coasian world often would not prevent the persistence of some inefficient structures that are already in place. Part III focuses on rule-driven path dependence. A country’s legal rules at any point in time, we argue, might be heavily influenced by the ownership patterns that 2 the country had earlier. We identify two reasons for the path dependence of rules— one grounded in efficiency and the other in interest group politics. First, even assuming that legal rules are chosen solely for efficiency reasons, the initial ownership patterns influence the relative efficiency of alternative corporate rules; the set of rules that would be efficient, we argue, might depend on the country’s existing pattern of corporate structures and institutions. 1 Stated formally, our claim is as follows. Let us denote by S1 and R1 the corporate structures and corporate rules that an economy has at time T1, and let us denote by S0 the structures that the economy had at an earlier time T0. Our claim is that S1 will be a function not only of R1, the legal rules prevailing in the economy, but also of S0, the corporate structures that the country had initially. 2 Stated formally, our claim is that R1 , the legal rules that an economy has at a given time T1, are a function of S0, the corporate structures that the economy had initially at T0. 2 Second, rule-driven path dependence might arise from interest group politics. A country’s initial pattern of corporate structures influences the power that various interest groups have in the process producing corporate rules. If the initial pattern provides one group of players with relatively more wealth and power, this group would have a better chance to have corporate rules that it favors down the road. Positional advantages inside firms will be translated into positional advantages in a country’s politics. And this effect on corporate rules will reinforce the initial patterns of ownership structure. For example, once a country has rules that favor professional managers and protect diffused ownership structures, these managers will have more political power and this power will in turn increase the likelihood that the country would continue to have such rules. Similarly, once a country has legal rules that enhance the private benefits to controlling shareholders and thus encourage the presence of such controllers, the controllers’ political power will also increase the likelihood that the country would continue to have such rules. To be sure, to the extent that a country has a suboptimal legal system due to interest group politics, this suboptimality might give incentives to those who set up companies to opt out of the country’s legal system through appropriate charter provisions or foreign incorporation or foreign listing. In a Coasian world, such mechanisms could lead to all companies being governed by the same efficient arrangements. As we explain, however, in an imperfectly Coasian world, these mechanisms are imperfect and cannot be expected to rigorously produce such a convergence. The focus of the analysis in Parts II and III is not on the possibility that corporate structures and corporate rules might be inefficient—but rather on the possibility that those structures and rules might be path dependent. Our analysis of path dependence differs from an analysis of possible inefficiencies in two ways. First, corporate structures and corporate rules can be both path dependent and efficient at the same time because, as we show, the identity of the efficient corporate structure or corporate rule might depend on a country’s original ownership patterns. Second, although another part of the analysis does concern the possibility that inefficient corporate structures or rules might arise, the focus of this part of the analysis is not on the possibility of inefficiency but on the role played by path dependence. Someone might accept that interest group politics can produce inefficient corporate rules but still expect roughly the same type of inefficient rules. For this reason, our analysis focuses not on the possibility that inefficient rules might arise but rather on showing why they would be likely to arise in different ways and to a different extent in different countries, depending on the countries’ initial conditions. For example, in our analysis of interest group politics, we focus on explaining why the inefficient legal rules resulting from interest group politics might vary among countries due to the initial patterns of corporate ownership structures. In Parts II and III, we will pay close attention to the forces created by increasing globalization. In both Parts, we will explain why the pressures exerted by global product and capital markets cannot be expected to eliminate path dependence. While we focus on path dependence, we also discuss in Part IV other reasons, not rooted in path dependence, why corporate structures might vary among countries 3 and continue to do so over time. Path dependence focuses on reasons why countries that are otherwise similar in all other aspects of their economy might still differ in their corporate structures. However, the advanced economies might differ in some relevant aspects. Differences in the nature of firms and markets, and in opinions, culture, ideology, and political orientation, might have all impeded, and might well continue to impede, convergence of corporate structures. Path dependence, then, can play an important role in the development of corporate ownership and governance structures around the world. The sources of path dependence that we identify can explain why (despite the powerful forces pressing toward convergence in an increasingly competitive and global marketplace) the advanced economies still differ in important ways in their patterns of corporate ownership and governance. The identified path dependence also indicates that some important differences might persist. I. EXPLAINING PERSISTENT DIFFERENCES In this Part, we describe our inquiry, define our terms, describe the competitive forces that could be seen as whittling away structural differences, and present the problem on which we focus: Why have different corporate structures persisted when so many other economic differences have not? We then identify two sources of path dependence that can help to answer this question. A. The Focus of Our Inquiry We focus on how countries differ in the structure of ownership and governance of their corporations—that is, how firms are owned and how authority is distributed among owners, the board of directors, senior managers, and employees. For concreteness, we focus on the relative dispersion or concentration of ownership of public companies. This dimension of corporate structure is important because the presence or absence of a controlling shareholder affects substantially the way in which, and the ends toward which, a corporation will be governed. At present, publicly traded companies in the United States and the United Kingdom commonly have dispersed ownership, whereas publicly traded companies in 3 other advanced economies commonly have a controlling shareholder. Indeed, while 3 For works comparing the incidence of controlling shareholders in different countries, see generally Marco Becht & Ailsa Roel, Blockholding in Europe: An International Comparison, 43 EUR. ECON. REV. 1049 (1999) (discussing the size of block shareholdings in Europe); Rafael La Porta, Florencio Lopez-de-Silanes & Andrei Shleifer, Corporate Ownership Around the World, 54 J. FIN. 471 (1999) (finding that only a few economies have many corporations that are widely held). For studies documenting the high incidence of controlling shareholders in specific European countries, see generally Luigi Zingales, The Value of the Voting Right: A Study of the Milan Stock Exchange Experience, 7 REV. FIN. STUD. 125, 131 (1994) (showing a high concentration of ownership in a sample of large public companies in Italy); Marcello Bianchi, Magda Bianco & Luca Enriques, Ownership, Pyramidal Groups and Separation Between Ownership and Control in Italy (Sept. 1997) (unpublished manuscript, on file with the authors) (finding a high concentration of ownership in Italy); Laurence Bloch & Elizabeth Kremp, Ownership and Control in France (Oct. 14, 1997) (unpublished manuscript, 4 most large American companies have diffuse ownership, eighty-five percent of the largest German firms persist in having a large shareholder (usually family, sometimes 4 financial) holding twenty-five percent or more of the firm’s voting stock. And while some observers believe that some “functional” corporate convergence has taken 5 place, there can be little doubt that, given the significance of controlling shareholders, countries that differ in their incidence of controlling shareholders have corporate structures that differ from each other substantially. These differences persist today despite the convergence of other economic institutions. We will also look at employee involvement in firms’ power structures. This is again an important dimension of current international differences. Labor is involved in the control of German corporations through codetermination, but does not have such direct, formal influence in corporations of other economies. Our focus will be on path-dependent bases for divergence. By path-dependent bases, we mean reasons arising from the different initial conditions with which countries started. Take two countries and assume that, while different in their initial corporate structures and legal rules, the two became identical some time ago in terms of their economies, politics, types of firms, cultures, norms, and ideologies. Could differences in corporate structures still persist? They could to the extent that a country’s corporate structures and rules depend, as we will argue, on the country’s initial corporate structures and rules. Given our interest in path dependence, we will focus on the corporate structures and rules prevailing in the world’s advanced economies. When two countries are at sharply differing levels of economic development, there would clearly be reasons other than path dependence for their ownership patterns to differ. We focus therefore on the advanced economies because their similar stage of economic development enables us to concentrate on path dependence. on file with the authors) (finding a high concentration of ownership in France); Julian Franks & Colin Mayer, Ownership, Control and the Performance of German Corporations (Jan. 25, 1997) (unpublished manuscript, on file with the authors) (finding a high concentration of ownership in Germany). 4 Franks & Mayer, supra note 3, at 8 (finding in a sample of 171 German firms that single owners held twenty-five percent or more of voting stock in eighty-five percent of these companies). 5 See, e.g., Steven N. Kaplan, Top Executives, Turnover, and Firm Performance in Germany, 10 J.L. ECON. & ORG. 142, 144 (1994) (finding analogous tendencies influencing turnover of board members in Japan, Germany, and the United States); Steven N. Kaplan & Bernadette A. Minton, Appointments of Outsiders to Japanese Boards: Determinants and Implications for Managers, 36 J. FIN. ECON. 225, 256-57 (1994) (suggesting that corporate governance in Japan plays essentially the same role as takeovers and proxy fights in the United States); Elisabeth Roman, Une nouvelle génération s’installe à la tête du capitalisme familial italien [A New Generation Sets up at the Head of Italian Capitalism], LE MONDE, May 15, 1998, at 16 (discussing how the new generation of Italian executives are increasingly following American business models); Greg Steinmetz, Changing Values: Satisfying Shareholders Is a Hot New Concept at Some German Firms, WALL ST. J., Mar. 6, 1996, available in 1996 WL-WSJ 3097228 (discussing a growing solicitude for shareholders by German executives). 5 B. The Persistence of Corporate Differences 1. Globalization and the drive toward efficient structures. It might be thought that the advanced economies should by now display similar patterns of corporate structure. Companies in these countries face similar governance problems. All large-scale firms share some key common functions: Capital must be gathered, management must be selected and disciplined, and information must be transmitted to core decisionmakers organizational imperatives could demand organizational similarities. And other powerful forces, it might be argued, drive countries and firms to adopt the most efficient corporate rules and structures. Not to do so in our competitive global village runs the risk that firms and the economy will fall behind. A firm that did not adopt the best structure would be hurt either in its profits and value or in its ability to raise new capital. Countries that fail to adopt efficient rules would inflict costs on their corporations, which would then be worth less and would then be less able to raise capital; as a result, firms, factories, and busi6 nesses might suffer, or they might migrate away from the country. Another way of stating the above view is that, as efficient new technologies can spread rapidly, one might expect (by analogy) that new corporate technologies, if better, should spread rapidly. Corporate governance could be seen as a technology— similar to a manufacturing technique, an inventory management system, or an engineering economy of scale—and firms face powerful incentives to adopt the best corporate technologies possible: The corporation and its securities are products in financial markets to as great an extent as the sewing machines or other things the firm makes. Just as the founders of a firm have incentives to make the kinds of sewing machines people want to buy, they have incentives to create the kind of firm, governance structure, and securities the customers in capital markets want.7 The adoption of the same efficient corporate governance technologies across the advanced economies might be facilitated, on the view under consideration, by the easy flow of information about corporate technologies. Cross-border investors and 8 multinationals bring with them familiarity with foreign practices. National reports 9 regularly consider practices seen elsewhere and identify them as beneficial. 6 See FRANK H. EASTERBROOK & DANIEL R. FISCHEL, THE ECONOMIC STRUCTURE OF CORPORATE LAW 212-18 (1991) (arguing that competition induces convergence in state rules); Roberta S. Karmel, Is It Time for a Federal Corporation Law?, 57 BROOK. L. REV. 55, 90 (1991) (“Despite these historical differences in corporate governance practice in the United Kingdom and continental countries, the laws will soon become more congruent . . . .”); cf. Harold Demsetz, The Structure of Ownership and the Theory of the Firm, 26 J.L. & ECON. 375, 375-77 (1983) (arguing that there is an ineluctable pressure on corporate structures toward efficiency). 7 EASTERBROOK & FISCHEL, supra note 6, at 4-5 (emphasis added). 8 In the late 1990s, this cross-border investment force tends to make corporate governance converge more towards American patterns than otherwise, because the international investors most active so far in pushing corporate governance initiatives have been Americans. See Martine Orange & Enguérand Renault, Les patrons français se sont convertis aux exigences des actionnaires [The 6 2. Persistence. Given these pressures to whittle down corporate differences, the question arises as to why corporate ownership and governance structures have continued to differ. To be sure, it is possible to point out movements that are reducing certain dif10 ferences—e.g., the efforts to encourage wider stock ownership in Europe, German 11 banks’ statements that they will sell off their stockholdings, the takeover headlines 12 in Europe, and the rising influence of American institutional investors in the United States (with the possibility that they will acquire the influence sometimes had by fi13 nancial institutions in continental Europe and Japan). But these stories are balanced by considerable persistence. For example, German banks, despite their rhetoric of withdrawal from stock ownership, have thus far held on to their stock. In fact, over the past decade, GerFrench Owners Have Converted to Stockholders’ Demands], LE MONDE, Apr. 23, 1998, at 21 (noting that foreign, typically Anglo-Saxon, stockholders in large French firms are pressing managers to be more concerned about shareholder value); John Tagliabue, Compliments of U.S. Investors: New Activism Shakes Europe’s Markets, N.Y. TIMES, Apr. 15, 1998, at D1 (noting spread of American shareholder activism); Sara Webb, Calpers Sees New Targets Overseas, WALL ST. J., Oct. 20, 1997, at C1, available in 1997 WL-WSJ 14170443 (noting Calpers’ attempts to improve the corporate governance of overseas companies). Daimler’s takeover of Chrysler might bring German governance practices in line with American ones (or further spread American governance practices in Germany). 9 See generally Corporate Governance Forum of Japan, Corporate Governance Principles—A Japanese View (Interim Report) (1997) (discussing in general corporate governance); Competitiveness Policy Council, Reports of the Subcouncils (1993) (discussing value of large institutional blockholding); PETER MÜLBERT, EMPFEHLEN SICH GESETZLICHE REGELUNGEN ZUR EINSCHRÄNKUNG DES EINFLUSSES DER KREDITINSTITUTE AUF AKTIENGESELLSCHAFTEN? [ARE RULES LIMITING BANK INFLUENCE DESIRABLE?] (1996) (discussing usefulness of alternative governance systems); ORGANIZATION FOR ECONOMIC COOPERATION AND DEVELOPMENT, BUSINESS SECTOR ADVISORY GROUP ON CORPORATE GOVERNANCE, INSTITUTIONAL MODERNISATION FOR EFFECTIVE AND ADAPTIVE CORPORATE GOVERNANCE: CHALLENGES AND RESPONSES (1997) (seeking basic worldwide principles of corporate governance); Michael E. Porter, Capital Disadvantage: America’s Failing Capital Investment System, HARV. BUS. REV., Sept.-Oct. 1992, at 65-82 (advocating adoption of foreign-based corporate governance innovations). 10 See Jeffrey N. Gordon, Pathways to Corporate Convergence? Two Steps on the Road to Shareholder Capitalism in Germany, 5 COLUM. J. EUR. L. 219, 220 (1999). 11 See, e.g., Brian Coleman & Dagmar Aalund, Deutsche Bank to Cash Out of Industrial Stakes, WALL ST. J., Dec. 16, 1998, at A17 (noting that Deutsche Bank announced plans to sell stock holdings); Role of the Financial Services Sector: Hearings Before the Task Force on the International Competitiveness of U.S. Financial Institutions of the Subcommittee on Financial Institutions Supervision, Regulation and Insurance of the House Committee on Banking, Finance and Urban Affairs, 101st Cong. 164-65 (1990) (noting that Deutsche Bank executive revealed intention to sell stock holdings). 12 See, e.g., Sophie Fay & Pascale Santi, L’offensive de la BNP plonge le monde bancaire dans la confusion [BNP’s offensive plunges the banking world into confusion], LE MONDE, Mar. 12, 1999, at 23 (noting that French bank made simultaneous hostile bids for two other large banks). 13 See Mark J. Roe, Strong Managers, Weak Owners: The Political Roots of American Corporate Finance 223-24 (1994) [hereinafter Roe, Strong Managers]. 7 many’s banks have increased the number of influential blocks they own in the one 14 hundred largest German firms from forty to over fifty. Thus, while German banks seem to have failed at their monitoring job in publicized cases, and while they have regularly been the target of populist sentiment, their considerable stock ownership has thus far persisted. Similarly, concentrated family ownership of Germany’s largest 15 firms persists. With respect to Japan, given the breakdown of the Japanese banking system, and the widespread recognition of the problems of Japanese corporate governance, one might have expected to observe a decline in banks’ ownership of large corporate blocks in Japan. Yet, the ownership data for the largest Japanese firms hardly indicate any movement in bank and insurer ownership in the largest firms over the past 16 three decades. In any event, it does not matter for our purposes whether the overall variance among countries in ownership structures has been recently narrowing somewhat, remaining the same, or increasing—a question which the data is insufficient to resolve. What is clear is that, notwithstanding the forces of globalization and efficiency, some key differences in corporate structures among countries have persisted. This observation raises important questions for researchers: Why have such differences persisted? And will they persist in the future? C. Sources of Path Dependence Our focus will be on the role that is played by path dependence in creating and maintaining differences in corporate structures. There are two sources of path dependence. One type of path dependence, which we will analyze in Part II, is structure driven. By structure-driven path dependence, we mean the ways in which initial ownership structures in an economy directly influence subsequent ownership structures. As we shall see, there are two ways through which an economy’s ownership structures might depend on its initial pattern of corporate ownership structures. The other type of path dependence arises from corporate rules. Such rules can influence corporate ownership and governance structures. In particular, such rules can shape choices between ownership structures that have and do not have a controlling shareholder. Corporate rules affect ownership and governance structures in at least three ways. 14 Compare Hauptgutachten der Monopolkommission, Marktöffnung umfassend verwirklichen [To Comprehensively Implement the Opening of the Market] 187-92 (1996/1997) (over fifty 5% or 5%+ financial institutional blocks in 1996), with Hauptgutachten der Monopolkommission, Wettbewerbspolitik oder Industriepolitik [Competition Policy or Industrial Policy] 205-12 (1990/1991) (about forty 5% or 5%+ institutional blocks in 1990). 15 See Franks & Mayer, supra note 3, at 25. 16 See generally MICHAEL S. GIBSON, “BIG BANG” DEREGULATION AND JAPANESE CORPORATE GOVERNANCE: A SURVEY OF THE ISSUES (Federal Reserve Int’l Fin. Discussion Paper No. 624, 1998) (concluding that reforms of financial institutions in Japan have thus far had a limited effect on corporate governance). 8 First, concentrated ownership might be discouraged by the presence of legal rules that make it more difficult or costly for financial institutions to accumulate and 17 hold large blocks. Such rules are strongly present in the United States, but not as 18 strong in other countries. Second, in corporate systems that enable controllers to extract large private benefits of control, “rent-protection” considerations might lead to concentrated own19 ership. When private control benefits are large, those who set up the corporate structure in an IPO would be reluctant to leave control up for grabs, because doing so would attract attempts to grab control and render the chosen structure unstable. Furthermore, when private benefits of control are large, controlling shareholders of publicly traded companies will be reluctant to relinquish their lock on control when raising extra capital, because they will not be compensated by existing shareholders for forgoing the larger benefits that come with a lock on control. Third, some countries have mandatory corporate rules that constrain, or push in a certain direction, the choice of governance structure. For example, some rules affect the constitution of the board of directors and the degree of labor influence in 20 the firm. American stock exchange rules and state corporate law doctrines militate in favor of a high proportion of independent directors. Japanese employee-oriented 21 norms lead to insiders dominating corporate boards. And German rules mandate 22 that labor has half of the board seats of large firms. Thus, given how important corporate rules are, substantial differences in such rules among countries might be sufficient to produce substantial differences in ownership patterns. Part III will focus on rule-driven path dependence. By rule-driven path dependence we mean the additional, indirect (but important) channel through which initial corporate structures might affect subsequent structures—by affecting future business rules. As we shall show, the corporate rules of an economy, which will have an effect on choices of ownership structures, are themselves influenced by the economy’s initial pattern of corporate structures. 17 See ROE, STRONG MANAGERS, supra note 13, at 26-49. See id. at 167-97. 19 See generally LUCIAN ARYE BEBCHUK, A RENT-PROTECTION THEORY OF CORPORATE OWNERSHIP AND CONTROL (National Bureau of Econ. Research Working Paper No. 7203, 1999) [hereinafter BEBCHUK, RENT-PROTECTION THEORY] (analyzing how rent-protection considerations affect the choice of ownership structure at the IPO stage); Lucian Arye Bebchuk, Rent Protection and the Evolution of a Firm’s Ownership Structure (July 1999) (unpublished manuscript, on file with authors) [hereinafter Bebchuk, Rent Protection and Evolution of Ownership Structures] (analyzing how rent-protection considerations influence choices of ownership structure after a firm goes public). 20 See Mark J. Roe, Political Preconditions to Separating Ownership from Control (September 1999) (unpublished manuscript, on file with authors) [hereinafter Roe, Political Preconditions]. 21 See Yasu Izumikawa, Amidst Calls for Corporate Governance Reform, Nissan Questions Role of Non-Executive Directors, IRRC CORP. GOVERNANCE BULL., July-Sept. 1997, at 21 (noting the Japanese view that nonexecutives contribute little to corporate governance). 22 See Katharina Pistor, Co-determination in Germany: A Sociopolitical Model with Governance Externalities, in EMPLOYEES’ ROLE IN CORPORATE GOVERNANCE (Margaret Blair & Mark J. Roe eds., forthcoming 1999). 18 9 II. STRUCTURE-DRIVEN PATH DEPENDENCE We begin our analysis of path dependence by analyzing structure-driven path dependence. We want to begin by focusing on the “direct” effect that the corporate structures in an economy at an earlier point in time have on structures at later points. Specifically, we show how an economy’s ownership structures depend on the pattern of ownership structures that the economy had at earlier points in time. Consider two advanced economies, A and B, which have at time T1 the same given set of legal rules and economic conditions but had earlier, at T0, different patterns of corporate ownership structures. Suppose, concretely, that at T0, companies in A commonly had a controlling shareholder and companies in B commonly have diffuse ownership. These structural differences at T0 might have been due to the countries’ having different legal rules or different economic conditions. While the two countries have reached T1 through different paths, at T1 they have the same corporate rules and economic conditions. Would these identical rules and conditions at T1 imply that the countries will also be the same from T1 on in terms of corporate structures? The answer is no. We will show in Part II.A how the initial pattern of ownership at T0 might affect the identity of the efficient structure for a given company at T1. We will then explain in Part II.B how internal rent-seeking behavior might provide existing corporate structures with some persistence power. A. Path Dependence of the Efficient Structure The first reason for structure-driven path dependence is grounded in efficiency. The identity of the efficient structure for a given company at T1 might depend on the earlier ownership patterns at T0 and might thus differ between A and B. This difference might be due to sunk adaptive costs, complementarities, network externalities, endowment effects, or multiple optima. We briefly explain each of these reasons. 1. Sunk adaptive costs. Sunk costs can influence the efficient choice of a corporate ownership structure. Consider the analogous situation in which maintaining an existing factory might be efficient even if a different factory would be more efficient to build if it were built from scratch: Once costs are sunk in equipment with no good alternative use, continuance often is efficient. In a similar way, sunk costs can be important for determining which corporate ownership structure might be efficient at a given point. For example, in a country in which diffuse ownership was common at T0, firms might have adapted by developing incentive compensation schemes for managers, by adding more independent directors, and by creating a debt structure that reduces agency 23 costs. Once such different adaptations take place at T0 in countries A and B (due to 23 Firms develop routines that give them a competitive advantage by lowering internal transaction costs. These embedded routines make a firm well adapted to its environment, but if the environment changes radically, the firm cannot easily unlearn its routines. It withers but does not adapt, and a 10 their different ownership structures at T0), these adaptations might make the efficient 24 ownership structure for a given company at T1 different in A and in B. 2. Complementarities. Complementarities are similar to sunk adaptive costs, but they concern adaptations not by the firm whose ownership structure is under consideration but rather by other entities and institutions. Institutions, practices, and professional communities often develop in every country to facilitate the working of the nation’s corporate structures. The corporate ownership structures that a country had earlier at T0 determined what accompanying institutions, practices, and skills were developed. And these aspects of the corporate environment might in turn influence what structures would be efficient later at T1. Suppose that diffuse ownership structures perform better in the presence of an active takeover market and transparent accounting, and that the development of such a takeover market and transparent accounting requires investments by firms and players to acquire the needed techniques and machinery. Whether a country had such activities developing at T0 would depend on what corporate structures it had back then at T0. In our example, such a market might have developed in country B in which diffuse ownership was common but not in country A in which diffuse ownership was rare. This implies that, for some firms, diffuse ownership might be efficient at T1 if they are in B but not if they are in A. 3. Network externalities. 25 Network externalities may also induce persistence. The efficient ownership structure for a given company might depend on the structures that other firms in the new firm arises with new but better-adapted routines. To the extent that this inability to unlearn embedded routines is true and applies to governance routines, adaptation is slow. See Rebecca M. Henderson & Kim B. Clark, Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms, 35 ADMIN. SCI. Q. 9, 9-10 (1990) (arguing that traditional categories of incremental and radical innovation are misleading); cf. Cristiano Antonelli, The Economics of Path-Dependence in Industrial Organization, 15 INT’L J. INDUS. ORG. 643, 644 (1997) (identifying switching and sunk costs as factors that induce irreversibility). To the extent that this potential rigidity of hardwiring is a problem, better governance will be more flexible governance. 24 One illustration: German firms probably adapted to codetermination by tending not to charge up their boardrooms, probably because neither managers nor shareholders were happy about enhancing labor’s voice in the codetermined boardroom. (German labor must get half of the supervisory board’s seats in the large firm.) They have used alternative governance structures to in-theboardroom governance: informal meetings between the management board and shareholders who own big blocks of stock. See Pistor, Co-determination in Germany, supra note 22; Mark J. Roe, German Securities Markets and German Codetermination, 98 COLUM. L. REV. 167, 168 (1998) [hereinafter Roe, German Codetermination]. Once the fit with codetermination was in place, the players may not have wanted to change ownership and governance. 25. Network externalities in corporate law are explored in Marcel Kahan & Michael Klausner, Path Dependence in Corporate Contracting: Increasing Returns, Herd Behavior and Cognitive Biases, 74 WASH. U. L.Q. 347, 350-53 (1996) (discussing learning and network externalities in corpo11 country have. There is an advantage to using the dominant form in the economy and the one with which players are most familiar. Thus, diffuse ownership may be less costly for a firm if other firms are diffusely owned. This consideration might make it efficient for a firm to choose a controlling shareholder structure if other firms in the economy commonly have such a structure—and choose a diffuse ownership structure if the firms in the economy commonly have such a structure. 4. Endowment effects. Endowment effects might also affect the identity of the efficient ownership structure. Players’ having control under an existing structure might affect their valuation of having such control, which would in turn affect the total value that alternative 26 structures would produce. To speculate, such an endowment effect might make it harder to transform both firms governed by European-style concentrated family owners and those governed by American-style managers. European family owners, being in control, might value their control highly. Similarly, American managers, already asset rich, might highly value their position and power. In either case, asking and offer prices might differ. Given the existing control structures, the value that these two groups attach to control is higher than what they would be willing to pay for it if they did not have it. In the presence of such an endowment effect, the overall efficiency of such control structures depends on whether they existed initially. 5. Multiple optima. Ownership structures affect corporate governance and corporate value in many complex ways. Thus, two alternative structures could each have pros and cons compared with the other, and they could thus produce roughly equal corporate value overall. Suppose that, under the corporate rules that countries A and B have at T1, concentrated ownership and diffuse ownership have largely offsetting pros and cons and thus that they are (roughly) equally efficient. Given that moving from one structure to another would involve transaction costs, maintaining the status quo might be efficient in each country. In this case, the initial pattern of corporate ownership in each of the economies can determine the subsequent pattern. * * * * Hence, sunk adaptive costs, complementarities, network externalities, endowment effects, and multiple optima might all lead the identity of the efficient ownership rate contracts), and Mark A. Lemley & David McGowan, Legal Implications of Network Economic Effects, 86 CAL. L. REV. 479, 562-86 (1998) (suggesting appropriate modifications of corporate rules based on network externalities). 26 To speculate on another possible endowment effect, it might be the case that German labor’s valuation of codetermination depends on the existing conditions. That is, it might be that German labor would demand more to give up codetermination than it would be willing to pay to get it in the first place. This implies that the overall efficiency of codetermination for a given German firm depends on whether or not it already has codetermination. 12 structure for companies at T1 to depend on the initial structure that the company and/or other companies in the economy had at T0. And this provides some reasons why the initial differences between countries A and B at T0 might persist later on at T1. B. Persistence of Existing Structures due to Rent-Seeking We now turn to the rent-seeking reasons for why structures that existed at T0 might have persistence power at T1. Due to rent-seeking, structures in place might be maintained even if they are no longer efficient at T1. Those parties that participate in control under an existing structure might have both the incentive and power to impede changes in the structure. Changing an ownership structure often requires the cooperation of those parties that control the firm. And the fact that a change in the ownership structure would be efficient would not ensure that controlling parties would always want it to occur. The controlling parties might prevent a change if it would reduce their private benefits of control whereas some of the efficiency gains would be captured not by them but by others. And in such situations, structures in place might per27 sist. 1. Persistence of concentrated ownership. Suppose that, under the legal rules that countries A and B now have, the efficient structure for a given company Y is diffuse ownership. If company Y had diffuse ownership to begin with at T0, then clearly it would continue to have diffuse ownership at T1. But suppose that Y is a company in country A and, like most other companies in country A, it began with a controlling shareholder. Y might not move at T1 to diffuse ownership. We next explain why. The controller’s roadblock. Suppose that Y has 100 shares, that at T0 an initial owner had all of the shares, and that at T0 she sold half of the shares to public investors and retained half of them as a control block. At T0, the initial owner had the incentive to choose the ownership structure that would maximize the value of the 100 shares, because at the time of decision, she owned all of the shares and internalized all of the effects of her decision. As such, we can suppose that concentrated ownership was the efficient structure at T0 given the conditions at the time and was therefore chosen at that time. By T1, however, the conditions have changed so that the total value of the company’s 100 shares would be higher under diffuse ownership than under concentrated ownership. Suppose that total value at T1 to all stockholders would be $100 in a concentrated structure—consisting of $60 to the controller ($1.20 per share in the control block) and $40 to the minority shareholders ($.80 per minority share). And 27 Cf. Stacey Kole & Kenneth Lehn, Deregulation, The Evolution of Corporate Governance Structure, and Survival, 87 AM. ECON. REV. PAPERS & PROC. 421 (1997). Kole and Lehn show that airline deregulation called for new governance structures for the airlines. Deregulation created more managerial complexity, calling for more incentive-based managerial pay, smaller boards, and more concentrated ownership. Incumbent firms adapted slowly, although new entrants entered the market with the superior governance structure in place. Evolution was, even after twenty years, incomplete. 13 suppose that the total value to stockholders would be $110 under diffuse ownership ($1.10 per share). Would the firm’s controlling shareholder elect at T1 to move to diffuse ownership? If the initial owner went public at the later time T1 (rather than earlier at T0), she would have clearly chosen diffuse ownership as it would produce the highest value. By selling all the shares to dispersed investors, she would have received $110. If she were to use a concentrated structure, then she would have received only $100: $40 for the shares she would have sold to dispersed public investors and $60 for the control block that she would have retained as controller (or, equivalently, $60 from the funds she would get by selling to someone else who would be the controller). Thus, choosing diffuse ownership in an IPO at T1 would have maximized the initial owner’s proceeds. Owning all 100 shares at T1, she would have chosen that structure which would have maximized their value, and under the new legal rules in T1, the value-maximizing structure would have been diffuse ownership, which the initial owner would have chosen. TABLE I DIVISION OF FIRM VALUE AT T1 UNDER CONCENTRATED OWNERSHIP Shares owned Fraction of Value owned Value Controllers’ block 50 shares 60% of value $1.20 per share Outsiders’ shares 50 shares 40% of value $.80 per share Total of firm 100 shares 100% of value $100 $ 60 in controller’s block + $ 40 in minority shares = $100 Total value 14 TABLE II DIVISION OF FIRM VALUE AT T1 UNDER DIFFUSE OWNERSHIP Shares owned Fraction of Value owned Value Controllers’ block 0 Outsiders’ shares 100 shares 100% of value $1.10 per share Total 100 shares 100% of value $110 $110 in cash Controllers’ total But because the company already went public at T0, at T1 it already has a concentrated ownership structure. So the question is whether the controller would move the firm toward diffuse ownership, a structure that would increase the firm’s total value by $10. It turns out that this “midstream” move to diffuse ownership would not 28 be in the controller’s interest. Consider the most straightforward route to accomplishing the change: the controller breaking up her control block and selling the shares in her control block to dispersed shareholders. Such a transaction would not benefit the controller. The total value of the firm under diffuse ownership is $110 (or $1.10 per share); thus the controller would receive only $55 from selling her remaining fifty percent of the company’s shares. That is, this sale would have provided the controller $5 less than the value of $60 that she would have by retaining her controlling block. Hence, she would not have benefited from breaking up her control block. To be sure, the move would raise the value of the shares that are already in the hands of public investors from $40 to $55, but this would not be a benefit that the controller would capture; the controller, of course, would not be able to raise retroactively the price at which the minority shares were sold from $40 to $55. Hence, the controller would not break up her control block at T1, and concentrated ownership would persist even though the move to diffuse ownership would increase total value. Alternatively, the move to diffuse ownership could take place at T1 if the controller would sell all the company’s assets to an entrepreneur and liquidate the company; the entrepreneur then would have the same incentives as an initial owner at T1 and those incentives would lead the entrepreneur to take the company public with dif28 For an analysis of analogous efficient structural changes that will not proceed due to similar roadblocks, see Lucian Arye Bebchuk, Efficient and Inefficient Sales of Corporate Control, 109 Q.J. ECON. 957 (1994) (analyzing how different legal rules governing the transfer of a controlling block might impede an efficient transfer); Mark J. Roe, The Voting Prohibition in Bond Workouts, 97 YALE L.J. 232, 277 (1987) (analyzing legally created obstacles for failed bond issues). 15 fuse ownership. But the most the controller would be able to get from the entrepreneur under this scenario would be $110 for all the assets, and the controller would receive in the subsequent liquidation only $55. This, again, would be less than the $60 in value that maintaining the control block would provide the entrepreneur. Under both of the considered scenarios, the controller would not benefit from the move to diffuse ownership because the move would eliminate the controller’s disproportionate access to the company’s value. Under the concentrated structure, the controller would capture sixty percent of the existing $100 pie, but a move to diffuse ownership would provide the controller with only fifty percent of the larger $110 pie. While the pie would grow larger, getting fifty percent of the somewhat larger pie would still be worse than getting sixty percent of the smaller pie under maintained concentrated ownership. Thus, even though the move to diffuse ownership would increase the firm’s total value by $10, the controller would not benefit from it; instead, she would lose $5. Another intuitive way for understanding why the controller would not benefit from the move to the more efficient structure is that the move would confer a positive benefit on the existing dispersed shareholders. The existing dispersed shareholders would end up with $55 if the controller moved to diffuse ownership instead of $40. This $15 benefit is one that the controller would not capture and thus would not internalize in her decisionmaking. Therefore, while the move would be efficient, the controller would not be served by it, because the controller would lose her rent (the private benefits of control) and would not fully capture the efficiency gains from the 29 move (some of which would be conferred on the existing pubic investors). In sum, whether or not the firm would have concentrated or diffuse ownership at T1 depends on its initial structure at T0. If the company were closely held at T0 and were to go public at T1, diffuse ownership would be chosen. Similarly, if the firm were to go public with diffuse ownership at T0, this structure would be maintained at T1. But if the firm went public with concentrated ownership at T0, this concentrated ownership would be retained at T1 and a move to diffuse ownership would not occur. Coasian alternatives? Might there be some other way in which the potential efficiency gain of $10 from the move to diffuse ownership could be realized? Would a gain of $10 be left on the table rather than taken? Couldn’t some transaction enable the parties to share the potential $10 gain? In a purely Coasian world, the players would indeed contract to implement the move and to realize and share among them this $10 gain. But in our imperfectly Coasian world, there are impediments to the realization of this $10 gain, and not all such gains will be realized. In a perfectly Coasian world, the move could take place through the minority shareholders’ paying the controller to induce her to move to diffuse ownership. Since the minority shareholders would gain $15 from such a move, and the controller would lose only $5 from such a move, a deal could benefit both sides. The minority could pay the controller some amount between $5 and $15, say $10, in return for the con29 The reason why the controller would not move to diffuse ownership is equivalent to the reason why a controller might not transfer control under an Equal Opportunity Rule even if the control transfer would be efficient. See Bebchuk, supra note 28, at 968-73. 16 troller’s agreeing to move to diffuse ownership. But in our imperfectly Coasian world, collective action problems among the minority shareholders would impede such a transaction. The shareholders would find it hard if not impossible to put together the “bribe” for the controller because of a “free-rider” problem. Each shareholder would know that her nonparticipation would barely affect whether the needed amount could be raised, and thus each would have an incentive to withhold her contribution. Alternatively, in a perfectly Coasian world, the controller could first buy the existing minority shares for $40, or for some amount between $40 and $50, and then move the firm to diffuse ownership and sell all of its shares for $110. As long as the payment to minority shareholders was below $50, the controller would in this way end up with more than the $60 that she would have had under concentrated ownership. But in an imperfectly Coasian world, the controller would find it difficult if not impossible to purchase the minority shares at such a price. Suppose that the controller were to make a tender offer for the minority shares at $.80 per share (or $40 in all). Such a tender offer might well fail due to a free-rider problem. Some public investors would be likely to hold out. A hold-out shareholder would see the value of her share go from $.80 to $1.10 if the other shareholders tendered and the controller thereafter moved the firm to diffuse ownership. And if all minority shareholders were to hold out for $1.10, the controller would not be able to buy the minority shares at a price 30 that would enable her to make any profit. The limits of persistence: large inefficiencies. Our argument is not that the move to diffuse ownership at T1 in the considered situation would fail no matter how large the potential efficiency gains. The move would take place if the potential efficiency gain were sufficiently large. Internal rent-seeking might enable a structure to persist only as long as its relative inefficiency is not too large. In the situation considered above, if the move would increase total value by more than $20—that is, if the value under diffuse ownership would be more than $120 at T1—then the controller would elect to move to diffuse ownership. Suppose that under diffuse ownership the total value of the firm at T1 would be $122. In this case, if the controller were to break up her control block and sell her shares to dispersed shareholders, she would receive $61, and this would give the controller more than the $60 that she would have had under concentrated ownership. The new pie of $122 would thus be enough to induce structural transformation. Even though the controller would still receive only fifty percent of this new pie, the new pie would be so large that this fifty percent would have a value larger than the sixty percent of the pie that she would have had under concentrated ownership. Our point is not that structures in place would persist due to rent-seeking no matter what. It is only that there is a wide range of values for which the controller’s rent-seeking would block an efficient move to diffuse ownership. In our example, as 30 In economic terms, there is no equilibrium in which the controller moves to diffuse ownership and prior to that pays the minority shareholders less than $1.10 per share. That is, there is no equilibrium in which the controller can benefit from acquiring all the shares through a tender offer and then transforming the firm to diffuse ownership. 17 long as the potential efficiency gains from a move (and thus the efficiency costs from maintaining the existing structure) are between $0 and $20, concentrated ownership would be maintained at T1. What determines the range within which concentrated ownership would persist even if it is inefficient? As the discussion of our example illustrates, the range depends on the size of the controller’s private benefits under concentrated ownership; the larger these private benefits, the larger the range in which an existing structure 31 will be maintained even if it ceases to be efficient. As long as these private benefits are significant at T1, this range of persistence will be significant in size. The limits of persistence: rent-destroying rules. The persistence of concentrated ownership that might result from rent-seeking would arise only if, under the legal rules at T1, controllers can enjoy rents in the form of some non-negligible private benefits of control. Thus, if countries were to adopt a legal regime eliminating such benefits altogether, this source of path dependence would be eliminated. Suppose that, at T1, the controller with fifty percent of the shares would capture no private benefits and thus get only $50, which is one-half of the pie under concentrated ownership. In this case, the controller would choose to move to diffuse ownership if and only if the move would increase total value. This qualification, however, would be relevant only under the unlikely scenario, which has not emerged yet, in which private benefits of control would not exist. New firms. The above analysis has focused on the persistence of structures in place. What about new assets that come into the economy and are put into corporate structures? Consider an economy populated at T1 by companies with concentrated ownership, and suppose that there are some resources owned by a sole owner at T1, and consider the choices that the owner will make for these assets. At this stage, since the sole owner has no partners, considerations of the owner’s internal rentseeking would not affect the choice of structure. However, the considerations identified in Part II.A. as to why the efficient structure might be path dependent—such as network externalities and complementarities—might affect the choice. Furthermore, the internal rent-seeking at work in other firms might influence where these assets end up. Controlling shareholders have an incentive to expand, because adding assets to their control will likely lead to increase in their private benefits of control. Consequently, in an economy populated by companies with a controlling shareholder, new resources will be often acquired by such companies even if these companies are not 32 the most efficient user of these assets. Whether for this reason or for the other rea31 Algebraically, if the fraction of the shares that are in the control block is k, the value under concentrated ownership is V and the private benefits of control are B, then the move to diffuse ownership will not take place as long as the value under diffuse ownership does not exceed V + [k/(1–k)]B. This condition can be derived in a similar way to the condition in Bebchuk, Efficient and Inefficient Sales, supra note 28, at 971, for when controllers will block efficient control transfers under the Equal Opportunity Rule. 32 See Bebchuk, Rent Protection and Evolution of Ownership Structures, supra note 19, at 1821 (analyzing how, other things equal, companies with a controlling shareholders will bid higher (than companies that are closely held) for assets that come into the corporate economy). 18 sons offered in this paper, observe that the flow of new assets and firms into the corporate sector has not thus far eliminated divergence. 2. Persistence of diffuse ownership. Diffuse ownership structures, once in place, might similarly persist due to internal rent-seeking by the incumbents managing such structures. Consider a company Y that, given the legal rules and conditions prevailing in countries A and B at T1, would produce the highest total value under concentrated ownership. Nonetheless, if the company’s initial structure at T0 was one of diffuse ownership, the firm might not move to concentrated ownership at T1. Suppose that Y has 100 shares; that its total value to shareholders at T1 under diffuse ownership would be $100 or $1.00 per share; that the managers would get control benefits of $3 under such diffuse ownership (from value diversions, prestige, etc.); and that under concentrated ownership the firm would produce a total value (to the controller and the minority shareholders combined) of $110 and a buyer is willing to pay this amount for the company in order to move it to concentrated ownership. While the move to concentrated ownership would be efficient, it might not take place. Notwithstanding that the move to concentrated ownership would increase total value, the existing managers might prefer that it not take place because it would eliminate their private benefits of control. And as long as the managers hold less than thirty percent of the shares, their fraction of the gains from the transformation would not be enough to make up for their loss of private benefits. The managers might be in a position to block or impede the move. They control the merger agenda and a merger cannot be initiated without their approval. They can also resist a hostile takeover bid. To be sure, if the potential gains from the move were very large, the move might still take place. But if the move would increase total value by ten percent, as in our example, and given the problems involved in a hostile bid, the managers might have not only the incentive but also the power to prevent the move. Thus, the desire of managers to keep the rents that they enjoy under the existing structure of diffuse ownership can provide such structure with some persistence power. Similar qualifications go with this conclusion as with the earlier conclusion concerning the possible persistence of concentrated ownership due to controllers’ rent-seeking. If the corporate rules at T1 provide the managers with no private benefits (a theoretical, unrealistic scenario because independence would always carry some benefits to the managers), then the managers would have no incentive to disfavor moves away from diffuse ownership. And if the legal rules at T1 give the managers no power with respect to acquisitions, then the managers would not have any power to resist a move. But as long as (i) managers derive some benefits from independence and (ii) managers have some power to resist acquisitions of control, then existing structures of diffuse ownership could have some persistence power. Thus, given that these conditions have been generally present in the past, this persistence might have played a role thus far—say, in maintaining such diffuse ownership structures in the United States— 19 even if a move to concentrated ownership could have increased value. And whenever these conditions will obtain in the future, this potential source of persistence and path dependence will remain relevant. 3. Persistence of German codetermination. Labor-preferring structures could persist for similar reasons. The most important example of a country in which labor participates in control is Germany. Germany has legal rules mandating labor participation in the board for all companies that are sufficiently large, and all such companies are thus codetermined. Our analysis suggests that dual-board structures might have some persistence power even if Germany’s legal rules change to make such structures optional rather than mandatory. Suppose that Germany changes its laws to make a dual-board structure optional rather than mandatory. Because dual-board structures are already in place, they might persist even if they are not efficient. If labor leaders (or other players) are getting private benefits from codetermination and if they have power to impede or resist 33 changes in the existing structure, they might resist a move away from codetermination. And as long as a Coasian bribe to labor leaders is illegal or transactionally costly, the move might not occur. We have now examined three principal “pure” types of firms, one with concentrated ownership, one with managerial control, and one with mandated labor influence. Each has a tendency to persist, and this persistence power contributes to a structural path dependence. 4. Persistence in the face of globalization. Thus, due to rent-seeking, structures in place could sometimes persist even if they cease to be efficient. A skeptic might question this conclusion, however, by wondering whether market forces in a global economy cannot always force controllers and managers to move to that structure that would be most efficient. But this is not the case. Our analysis already took into account whatever effects that might arise from product market and capital market competition. When we said that the firm’s value in our examples at T1 would be $100 under the existing suboptimal ownership structure and $110 following a move to a superior ownership structure, this difference of $10 incorporated already all the effects on total value from all potential sources, including product and capital market competition. And we have shown is that such a difference in total value might be insufficient to induce parties in control to favor the move to the superior structure. To be sure, globalization would discourage persistence of a suboptimal structure if the difference in total value between the best structure and the suboptimal one 33 We assume here that the German reform would track the standard American practice, with a firm’s governance changes being initiated by the board. But even if shareholders can initiate a repeal of codetermination, the result might not differ if shareholders concluded that the shock to labor of throwing them off the board would lead to unrest or demoralization. 20 is large enough. That is, globalization would end persistence if inefficiencies are always so large that they would largely obliterate firms with suboptimal structure, i.e., that there would be no “mere ten percent” inefficiencies. But even with strong global capital and product market competition, not every inefficiency in structure would have such drastic consequences. Even with globalization, an existing structure could have some limited (rather than unbounded) efficiency costs (say, ten percent of total value as in the examples we used) and thus would have some persistence power. Product market competition. To examine the above point in more detail, let us consider why product market competition, whether domestic or global, would not always be sufficient to prevent controllers or managers from sticking to an inefficient 34 structure. While maintaining a corporate structure might involve some efficiency costs and reduce shareholder value, it would not necessarily render the company unable to compete in its product market. While product market competition gives controllers, managers, and labor leaders valuable incentives for efficiency, it cannot always discourage them from main35 taining a structure that yields them private benefits but is somewhat inefficient. For one thing, a firm’s choice between concentrated ownership and diffuse ownership need not affect the firm’s costs or the quality of its products; rather, it might alter how the shareholders, managers, and controllers divide up the value produced by the firm. When a company’s ownership structure does not affect product quality or costs, product market competition will not constrain the company’s choice of ownership structure. Even if the choice of ownership structure affects the operational efficiency of a firm, product market competition often constrains the firm and its managers only 36 weakly. Product markets are not always perfectly competitive. Oligopolies can create slack, and managers and controllers can take advantage of it. Because product market competition does not threaten firms’ survival in such markets even if the firms forego some efficiencies, controllers and managers might sacrifice some potential efficiencies for the private benefits that maintaining the existing structure would yield. Global capital markets. The world’s ever-more-global capital markets provide firms, it might be argued, with incentives to adopt efficient ownership structures. If a firm maintains an inefficient structure, so the argument goes, the firm would be penal34 Cf. Frank H. Easterbrook, Managers’ Discretion and Investors’ Welfare: Theories and Evidence, 9 DEL. J. CORP. L. 540, 557 (1984) (arguing that the product market constrains managers and controllers to choose efficient structures and arrangements). 35 Cf. Lucian Arye Bebchuk, Federalism and the Corporation: The Desirable Limits on State Competition in Corporate Law, 105 HARV. L. REV. 1435, 1466 (1992) [hereinafter Bebchuk, Federalism] (analyzing why product market competition “cannot discourage managers from seeking valuedecreasing rules that are significantly redistributive in their favor”); Lucian Arye Bebchuk, Limiting Contractual Freedom in Corporate Law: The Desirable Constraints on Charter Amendments, 102 HARV. L. REV. 1820, 1845-46 (1989) [hereinafter Bebchuk, Limiting Contractual Freedom] (discussing how product market competition cannot discourage managers from seeking value-decreasing charter amendments that are significantly redistributive in their favor). 36 See JEAN TIROLE, THE THEORY OF INDUSTRIAL ORGANIZATION 277-303 (1988) (discussing imperfect competition). 21 37 ized in the capital markets and would face hurdles in raising new capital. But would globalized capital providers really strike down inefficiently governed firms by refusing to finance the firms’ futures? Global capital markets cannot generally be relied on to press managers to 38 move to the most efficient ownership structure. Many established companies do not use capital markets for funds, but rather finance themselves from retained earnings. When firms do not rely on external finance, their managers and controllers will not be constrained by capital markets. Among companies that do use external finance, some use debt rather than equity, and debt markets might not often constrain a structural choice because the structural choice might have little effect on the likelihood that the company will default on its debt. Indeed, even for firms that finance themselves by raising equity, the strength of the capital market constraint is uncertain. An inefficient ownership structure might merely mean that the company would have to issue more shares to raise a given amount of capital. This might not seriously discourage professional managers from inefficiently maintaining a diffuse ownership structure (if they own little equity themselves). And while it might somewhat constrain controllers (who would be diluting their own holdings by issuing more shares), even they might elect to maintain the existing structure and absorb such dilution for a time when raising equity if their private benefits of control under the existing structure are large enough. Thus, while there are limits here, inefficient structures might persist in the face of globalized capital markets. C. Conclusion on Structure-Driven Path Dependence The ownership structures that an economy has partly depend on the ownership structures that the economy had earlier on. Even if two nations have identical corporate rules and economic conditions at T1, if their initial structures differed at T0 (due to earlier differing economic conditions, for example), these differing structures at T0 could lead to differing structures at T1. There are two main sources for this structuredriven path dependence. First, the original structures affect which structure will be efficient for any given company: Sunk adaptive costs, complementarities, network externalities, endowment effects, and multiple optima might all make the identity of the efficient ownership structure depend on earlier structures. Second, initial structures might persist because players that enjoy rents under them might have both the incentive and power to impede changes in these structures. These two sources of structure-driven path dependence can help explain some key differences in ownership structures among the advanced economies that have persisted thus far. This structural 37 See Easterbrook, supra note 34, at 557. Cf. Bebchuk, Federalism, supra note 35, at 1465-66 (analyzing how capital market constraints cannot generally constrain managers from seeking some inefficient state law rules that favor them); Bebchuk, Limiting Contractual Freedom, supra note 35, at 1844-45 (analyzing how capital market constraints cannot generally discourage managers from seeking inefficient charter amendments). 38 22 path dependence might also lead to important differences among countries’ corporate structures in the future. III. RULE-DRIVEN PATH DEPENDENCE 39 Corporate rules can affect corporate governance. Thus, when two countries’ corporate rules differ, this difference by itself might produce differences in their patterns of corporate ownership structures. This raises the questions of why—given that the advanced economies all have an interest in providing their companies with desirable corporate rules—their systems of corporate rules have been so different and whether they will continue to differ in the future. Corporate rules, we argue, are themselves path dependent. The rules that an economy has at any given point in time depend on, and reflect, the ownership and governance structures that the economy had initially. This provides another channel through which initial ownership structures can affect subsequent choices of structure: The initial structures affect future corporate rules which in turn affect future decisions on corporate structures. Consider two economies that have similar economic conditions at T1. As we explain below, the corporate rules that A and B have at T1 might depend on the ownership structures (and thus also on the rules) that A and B had earlier at T0. That is, if A and B had different patterns of ownership structure at T0, their rules at T1 might well differ as a consequence. We observe in Part III.A that differences among systems of corporate rules should be assessed not by looking at general principles but rather by examining all aspects of the corporate rules system, including elements of procedure, implementation, and enforcement. In Parts III.B and III.C we identify and analyze two sources for the path dependence of corporate rules. We first show (Part III.B) how the preceding conditions of an economy at T0 might affect the choice of corporate rules at T1 even assuming that lawmaking is solely public regarding; this might result because the initial pattern of ownership might affect which legal rules would be efficient. We then show (Part III.C) how path dependence might arise when lawmaking is also influenced by interest group politics. In this case, the initial pattern of ownership might influence the relative political strength of various groups of corporate players. Both of these sources of rule-driven path dependence, as we will see, might often reinforce existing patterns of ownership. And they both might help explain why, even though the advanced economies have converged along many economic dimensions, their systems of corporate rules differ so much. A. Systems of Corporate Rules We first should clarify what we mean by saying that two countries have different corporate rules or different systems of corporate rules. General principles of cor39 See text accompanying notes 17-19 supra. 23 40 porate law may often be the same across countries, but more is at stake. Thus, all advanced countries may recognize and accept a certain fiduciary principle, but coun41 tries A and B might implement it radically differently. Principles are important, but “the devil is in the details,” and implementation counts a great deal. Two countries may be hostile to self-dealing in principle, but their overall legal treatment of selfdealing might differ greatly because of differences in the procedures that corporations must follow in approving a self-dealing transaction, in the nature and timing of the disclosures that the firm or the controller must make, in the incentives that public investors or plaintiffs’ lawyers have to sue, in the procedures that such suits have to follow, in the standards of scrutiny that courts use, in the level of deference that courts give to the insiders’ judgments, in the extent to which an effective discovery process is available, and in the ways in which evidence will be brought and considered. What counts are all elements of a corporate legal system that bear on corporate decisions and the distribution of value: not just general principles, but also all the particular rules implementing them; not just substantive rules, but also procedural rules, judicial practices, institutional and procedural infrastructure, and enforcement capabilities. Because our concern is with the corporate rules system “in action” rather than “on the books,” all these elements are quite important. Finally, in assessing the scope of the corporate rules system, recall that by corporate rules we mean throughout all the rules that govern the relations between the corporation and all of its investors, stakeholders, and managers, as well as among these players. Thus, for the purposes of our analysis, the corporate rules system includes not only the rules of corporate law as conventionally defined but also securities law and the relevant parts of the law governing insolvency, labor relations, and financial institutions. B. Path Dependence of the Efficient Rules Suppose that lawmakers in a given country are completely public regarding. Even so, rules might be path dependent because the identity of the locally efficient legal rule—the rule efficient for a given country—might depend on the rules and structures that the country had at earlier times. 1. Sunk costs and complementarities. Sunk costs and complementarities can induce efficient persistence. Different sets of rules might be more suitable for different types of companies. Public- 40 See Henry Hansmann & Reinier Kraakman, The End of History for Corporate Law, in ARE CORPORATE GOVERNANCE SYSTEMS CONVERGING? (Jeffrey Gordon & Mark J. Roe, eds., forthcoming 2000) (noting that the basic law of corporate governance has achieved a high degree of uniformity across Europe, America, and Japan). 41 Cf. Gérard Hertig, Convergence of Substantive Rules and Convergence of Enforcement: Correlation and Tradeoffs, in ARE CORPORATE GOVERNANCE SYSTEMS CONVERGING? (Jeffrey Gordon Mark J. Roe, eds., forthcoming 2000) (describing variance in the quality of enforcement). 24 regarding public officials might choose at T1 those rules that are best taking into ac42 count the structures and rules that were in place at T0. Existing legal rules might have an efficiency advantage because institutions and structures might have already developed to address needs and problems arising under these rules. In such a case, replacing the existing rules might make the existing institutional and professional infrastructure obsolete or ill fitting and require new investments. Various players—managers, owners, lawyers, accountants, and so forth— might have invested in human capital and modes of operation that fit the existing corporate rules. Replacing these rules would require these players to make new investments and to adapt to the new rules. Thus, which rules might be efficient for a country at T1 might depend on which rules it had at T0 and what institutions and practices developed in reaction to these rules. Note that this factor would often reinforce existing rules and, in turn, existing ownership structures. 2. Multiple optima. The path dependence of the rules that would be efficient for a given economy might also result from multiple optima. Suppose that technologically identical firms exist at T1 in countries A and B. Suppose that, at T0, A’s corporate rules favored concentrated ownership and A’s firms commonly had concentrated ownership. And suppose that B’s rules at T0 favored diffuse ownership and its firms commonly had such a structure. Suppose that, while the types of inefficiencies prevailing in A and B might well differ, both A’s rules and B’s rules (and in turn A’s structures and B’s structures) have aggregate costs of similar magnitudes. In this case, even assuming that public officials are completely public regarding in both A and B, neither set of officials would see a reason to switch (and, given the costs that would be involved in making changes, would thus see a reason not to switch) to the other country’s rules. C. Path Dependence of the Rules that Are Actually Chosen Law is of course not always made by public-regarding officials uninfluenced by interest groups. Interest groups might influence the choice of legal rules, which might sometimes lead to inefficient rules being chosen or maintained. The dynamics of interest group politics depends on the existing pattern of corporate ownership. This 42 Why wouldn’t each country adopt two separate bodies of corporate rules, one for companies with concentrated ownership and one for companies with dispersed ownership? Although different governing rules are possible, countries generally have one body of corporate rules, presumably because of the economies of scale involved in having one body of law and the problems resulting from (1) the need to decide which body of rules to apply and (2) players’ trying to manipulate their classifications. Those familiar with the history of American corporate bankruptcy might recall the unsuccessful experience of the Chandler Act, in force in the United States from 1938 to 1978. The Chandler Act provided one set of rules for public companies (Chapter X), another for privately held firms (Chapter XI). However, in the later stages of the act’s history, public firms tried, often successfully, to use the set of rules intended for nonpublic firms. In 1978, Congress felt compelled to abandon the two separate systems. See H.R. REP. 95-595 (1977), reprinted in 1978 U.S.C.C.A.N. 5963. 25 introduces another source for the path dependence of legal rules which we next exam43 ine. 1. Initial conditions and the political economy of corporate rules. Legal rules are often the product of political processes, which combine publicregarding features with interest group politics. To the extent that interest groups play a role, each interest group will seek to push for rules that favor it. Thus, the corporate rules that actually will be chosen and maintained might depend on the relative strength of the relevant interest groups. Interest groups differ in their ability to mobilize and then exert pressure in favor of legal rules that favor them or against rules that disfavor them. The more resources and power a group has, the more influence the group will tend to have in the political process. This is the reason why interest group politics might be influenced 44 by the existing distribution of wealth and power. In particular, the existing corporate ownership structures will affect the resources (and hence political influence) that various players will have and thus the rules that will be chosen. Hence, corporate rules at each point in time will depend on the economy’s existing corporate structures at earlier points in time. This path dependence will often induce bodies of corporate rules to differ among countries. When a certain set of rules leads corporate control to be at the hands one group of players, their control of existing structures will make these players more influential in subsequent interest group politics and will thus make it more likely that the country will have these or similar rules in the future. Their power within corporations will translate into power in the political process and influence on corporate rules. 2. Rules affecting concentrated and diffuse ownership structures. The legal rules favoring concentration or dispersion of corporate ownership affect corporate players, and these players might be influential interest groups. The 43 Several works in progress develop political economy explanations as to why the corporate rules of countries might differ. See generally Marco Pagano & Paolo Volpin, The Political Economy of Corporate Governance (1999) (unpublished manuscript, on file with authors); Raghuram G. Rajan & Luigi Zingales, The Politics of Financial Development (Aug. 1999) (unpublished manuscript, on file with the authors) (discussing the effect of politics on a country’s financial development); Roe, Political Preconditions, supra note 20 (stressing that how the agency costs of concentrated and dispersed ownership compare might differ from country to country due to political and cultural factors). The explanations in these papers focus on differences among countries in the political processes and underlying conditions rather than, as we do here, on how the very existence of interest group politics introduces path dependence. 44 See generally MAXIM BOYCKO, ANDREI SHLEIFER & ROBERT VISHNY, PRIVATIZING RUSSIA (1995) (discussing the effects of the initial distribution of property rights emerging out of privatization on the subsequent interest group politics); Jonathan R. Hay, Andrei Shleifer & Robert W. Vishny, Toward a Theory of Legal Reform, 40 EUR. ECON. REV. 559 (1996) (arguing that legal rules should accommodate rather than interfere with existing business practice). 26 power of controlling shareholders and of professional managers—and how influential they will be in corporate law politics—clearly depends on the existing ownership structures. Thus, the likelihood that rules favored by these groups will be chosen or maintained at any point in time will depend on the power that these groups have under the existing pattern of ownership structures. Consider anti-takeover rules that discourage the hostile acquisition of a company with a diffuse ownership structure. In the United States, there is an arsenal of 45 such laws, both statutory and judge made. Such rules encourage diffuse ownership and are beneficial to the professional managers of such companies. Now, a country that has mostly diffuse ownership to begin with would have more interest group support for such rules than one without diffuse ownership to begin with. Professional 46 managers benefit from such rules, and they can use corporate resources to lobby 47 lawmakers. And professional managers are clearly a much more powerful group in a country with diffuse ownership (such as the United States) than in one with concentrated ownership (such as Germany). Thus, a country that has more companies with diffuse ownership to begin with also would be more likely to have down the road anti-takeover rules—rules that might reinforce the tendency toward diffuse ownership 48 structures. Another example of rules that are more likely to be adopted or maintained in a country with diffuse ownership are rules discouraging financial institutions from ac49 tively acquiring and using large blocks of stock. Professional managers of companies with diffuse ownership favor such rules and have lobbied for them in the United States. The more powerful such managers are at any point in time, the more likely such rules will be adopted or maintained. Thus, a country that has diffuse ownership at T0 (with or without such rules) is more likely to have such rules adopted or main45 See generally RONALD GILSON & BERNARD BLACK, THE LAW AND FINANCE OF COR(2d ed. 1995) (surveying the legal rules governing takeovers). 46 Controlling shareholders are less interested in anti-takeover rules, because a controlling shareholder with enough shares can stop a hostile takeover by itself, without any help from antitakeover rules. 47 See generally C. Edwin Baker, Realizing Self-Realization: Corporate Political Expenditures and Redish’s The Value of Free Speech, 130 U. PA. L. REV. 646 (1982) (exploring the market forces that dictate the content of commercial speech); Victor Brudney, Business Corporations and Stockholders’ Rights Under the First Amendment, 91 YALE L.J. 235 (1981) (discussing the First Amendment contours of corporate speech). 48 For analyses of how American managers have obtained a body of takeover law that increasingly makes hostile takeovers difficult, see Lucian Arye Bebchuk & Allen Ferrell, Federalism and Takeover Law, 99 COLUM. L. REV. 1168 (1999); Mark J. Roe, Takeover Politics, in THE DEAL DECADE 321 (Margaret Blair, ed. 1993). 49 Such rules, which exist in the United States but not to the same extent in other advanced economies, discourage institutional ownership and thereby increase dispersed ownership. Attempts to reform many antiquated American financial rules have proved difficult and have proceeded slowly. See ROE, supra note 13, at 100, 229. For a description of a recent failure of such reform, see Richard W. Stevenson, House Leaves Finance Law of 30’s Intact: Bank Lobbying Delays Glass-Steal Repeal, N.Y. TIMES, Apr. 1, 1998, at C1 (noting how intense bank lobbying prevented repeal of a depressionera law). PORATE ACQUISITIONS 27 tained at T1 —and such rules will make it more likely that the initial incidence of diffuse ownership will be maintained or even increased at T1. Let us now turn to legal rules that are more likely to arise when ownership is concentrated and to further reinforce the prevalence of concentrated ownership. Rules that enable controllers to extract large private benefits of control are beneficial to controllers of existing publicly traded companies. In a country in which ownership is largely concentrated at T0 (with or without such rules), controlling shareholders of existing companies will be a powerful interest group with substantial resources. The influence of this group will make it more likely that this country will have or maintain 50 such rules at T1. And because such rules encourage the use or retention of concen51 trated ownership, the presence of such rules at T1 will in turn help maintain or even strengthen the initial dominance of concentrated ownership. Thus, control over corporate decisionmaking and resources also provides political power. Those who have on-share corporate control—be they controlling shareholders, professional managers, or other players—are likely to have influence because of the resources that they command. These resources will enable them to lobby, make campaign contributions, and otherwise gain political influence. These resources also provide them with visibility, access to media, high social status, and access to elite and influential groups, all of which can be helpful in influencing the corporate rules system. The fact that those in control of corporations can push to retain or expand legal rules that favor them might move path dependence, as we have seen, in a direction reinforcing existing ownership patterns. This might occur when professional managers in diffuse ownership countries support anti-takeover rules or rules discouraging financial institutions from holding blocks, and when controlling shareholders in concentrated ownership countries support rules that yield them large private benefits of control. Such an analysis might apply as well to rules establishing labor-preferring 52 structures, such as German codetermination. 50 Changes in corporate law generally apply to both existing and future companies. This feature tends to make existing rules persist. If it were otherwise, controlling shareholders might be indifferent to rules that would prevent future and new controlling shareholders from diverting value, as long as they, the incumbent controllers, were governed by the old rules that enable them to divert. See David Charny, The Politics of Corporate Convergence, in ARE CORPORATE GOVERNANCE SYSTEMS CONVERGING?, supra note 40 [hereinafter Charny, Politics]. But this dichotomy would be hard to argue for convincingly, hard to enact and hard to enforce. Interest groups usually must present principled positions, then push for what they term a principled view. For a discussion of how the presence of controlling shareholders might impede corporate reforms aimed at reducing private benefits of control, see generally Bebchuk, Rent Protection and Evolution of Ownership Structures, supra note 19, at 25-26. 51 See generally BEBCHUK, RENT-PROTECTION THEORY, supra note 19; Bebchuk, Rent Protection and Evolution of Ownership Structures, supra note 19. 52 Once codetermination is in place, labor leaders have more power. And to the extent that these leaders benefit from codetermination, their greater power in the system’s initial conditions will increase the chances that codetermination will persist. Employees may also have resisted changes and have had the votes to succeed. See generally Pistor, supra note 22, at 163 (showing resistance to changing German codetermination). 28 3. Globalization and the pressure to adopt efficient rules. A possible objection to the above analysis is again one based on globalization. Increasing globalization should discourage countries from ever adopting inefficient corporate rules, so the argument goes, because the economies of those countries that do so would suffer. Globalization, however, has not thus far had this effect, which is indeed far from surprising. Countries can preserve inefficient rules and can do so for long periods of time. There is in fact no mechanism that ensures that political processes will 53 only produce and retain efficient arrangements. Suppose that a country’s legal rules favor an outmoded governance system. Must the rules or its constituent firms have collapsed under the threat of heightened international competition? Is it unstable? The answer is no. What counts is whether the firms produce competitive products that can be sold. The firm can compete, even with an outmoded governance structure, if it makes up for this governance disadvantage with an offsetting international competitive advantage. If the firm can pay for an immobile input at a lower price than firms in other countries can, it can readily survive. Or, the country might subsidize the firm (directly or via lower taxes) with higher taxes elsewhere (on an immobile element of the economy). This result, while reducing that nation’s standard of living relative to others’ (and accordingly, it has some limit), does not necessarily lead to economic instability. Stability depends as 54 much on a nation’s politics as it does on global competition. Interest group politics can lead countries to inefficient arrangements. Globalized capital and product markets impose costs on firms laboring under inefficient legal rules, but if a country is prepared to bear those costs, or if positionally powerful players inside the firm can make those costs be borne by outsiders, even outmoded and costly rules can persist. 4. Can contracts generally substitute for legal rules? A critic might argue that, when a country chooses inefficient legal rules, corporate players will avoid them by adopting efficient arrangements through contracts. While we agree that contracting around inefficient rules can often work, it cannot generally do so. Mandatory rules often make contracting around impossible. And 55 even when contracting around is allowed, it is often too costly to do so. 53 See MANCUR OLSON, THE RISE AND DECLINE OF NATIONS: ECONOMIC GROWTH, STAGSOCIAL RIGIDITIES 17-35 (1982) (arguing that established groups impede change). Countries with inefficient legal rules might not even suffer an aggregate disadvantage, if all countries have some inefficient rules. All countries could have inefficient legal rules, but their rules might be inefficient in different ways with the differences being partly path dependent. 54. See Mark J. Roe, Backlash, 98 COLUM. L. REV. 217, 219-21 (1998). 55 Norms also cannot easily substitute for legal rules, for reasons similar to those that impede contracts from substituting for rules. See David Charny, Nonlegal Sanctions in Commercial Relationships, 104 HARV. L. REV. 373, 429-44 (1990); Charny, Politics, supra note 50; Eric A. Posner, Law, Economics, and Inefficient Norms, 144 U. PA. L. REV. 1697, 1728-36 (1996). FLATION, AND 29 True, some rules are technical, involving only two parties, and can easily be reversed by contract. If the “default” rules favor managers (or controllers) and the parties can change the corporate charter, they sometimes may do so. But three simple examples make the point that there are limits. First, one nation may induce, intentionally or not, diffuse ownership by keeping capital-gathering institutions small and barring them from actively owning large blocks of stock; those who want to contract around these rules would have to build a parallel, unregulated financial system—a costly, perhaps impossible task—and the forces that made the first system illegal will 56 likely make the second one illegal as well. Second, consider a nation’s failure to reduce the benefits that a controlling shareholder can extract from a firm. The corporate charter, or contract, could take the extraction-reducing rules that another nation has and impose them on the controlling shareholder. But adopting these rules in the corporate charter might provide limited benefits if, to be effective, the rules need the implementation system—the courts, precedents, professionals, and norms—that the other nation has. This implementation system is a “public good” as to the contracting parties and cannot be readily built by those parties to the two-way contract. Third, for an example of mandatory rules that cannot be readily contracted around, consider our example of the German rules requiring codetermination, which mandate that half of the firm’s supervisory board be labor representatives. There is no formal way to contract around this rule. A parallel structure would lack formal authority inside the firm and, if given formal authority (as was occasionally attempted 57 in Germany in the 1970s and 1980s) would be illegal under German law. 5. Reincorporations. Another way for corporate players to “contract around” an inefficient system of corporate rules is by reincorporating in another country. For example, a foreign firm can subject itself to U.S. rules by reincorporating as, say, a Delaware corporation, or it can subject itself to some subset of U.S. rules by selling shares in the United States. Reincorporation could, in theory, enable each company to remove itself from the local interest group politics (if local rules are inferior to those of some other country) and to get the rules of that other country by reincorporation. With costless reincorporation, firms could migrate to those countries with the most attractive legal 56 Cf. ROE, STRONG MANAGERS, supra note 13, at 60-93. American financial law barred interstate banking and banks with big blocks of stock at the end of the nineteenth century. American life insurers tried to end run this bar by building an interstate insurance system, with the insurers owning big blocks of stock. But by 1906, new law barred the insurers from active ownership of large blocks of stock. Other substitutes besides a parallel financial system are imaginable, but they also may be too costly or might be barred. Cf. Barry E. Adler, Politics and Virtual Owners of the Corporation, 82 VA. L. REV. 1347, 1362-64 (1996). 57 German courts struck down efforts to contract around codetermination by using subcommittees having a reduced labor representation. See Roe, German Codetermination, supra note 24, at 168. 30 rules, with a resulting pressure on countries to adopt efficient rules lest they lose all incorporations to other countries. The possibility of reincorporation has indeed profoundly affected corporate rules in the United States. Reincorporations have led to the migration of many firms to Delaware and to the adoption by many states of rules that approximate the rules 58 prevailing in Delaware. But these migrations have been facilitated in the United States by the fact that American companies are treated similarly throughout the country irrespective of the state in which they have been incorporated. Consequently, for an American company to reincorporate from one state to another is a “pure” choice of a corporate law system and involves no other economic consequences. This, however, is not the case in today’s world for reincorporations from one country to another. Such reincorporations cannot be made simply as an instrument for choosing a different set of corporate rules because they will usually carry with them significant tax, regulatory, or other economic consequences. And as long as such impediments to reincorporation exist, reincorporations cannot replicate at a world level the effect that they have had on corporate rules in the United States. The above discussion suggests a caveat. If the world had moved to one big federal system, then differences among countries in their corporate rules would have largely disappeared or receded. But this worldwide federal system has not emerged thus far. Steps in this direction have been tentative and infrequent. And as long as it does not emerge, the source of path dependence that we have identified in this Part will continue to operate. Moreover, even if reincorporations in another country were costless, they would enable firms and corporate players to avoid only those corporate rules that depend on the place of incorporation. But the system of corporate rules governing the relations between the corporation and its stakeholders also includes many elements which do not depend on the place of incorporation— such as the rules governing insolvency, banking, or labor contracts—and which thus cannot be avoided by reincorporation. 6. Public-regarding victories over interest group politics. While we have focused in this Part on interest group politics, we do not assume that corporate rules are solely the product of interest group pressures. As we have shown in Part III.B, corporate rules will be path dependent even assuming that lawmakers are completely public regarding. Our goal in this Part has been to show that, to the extent that interest groups play a role, and people might reasonably disagree on how substantial a role they do in fact play, this role will depend on existing ownership structures. Below we offer some remarks on how efficient corporate rules might be adopted despite interest group politics—and point out that the identity of the 58 Academic disagreements persist regarding whether the competition among states in the U.S. has been beneficial. Compare Bebchuk, Federalism, supra note 35 (analyzing the problems with competition among states over corporate incorporations), with ROBERTA ROMANO, THE GENIUS OF AMERICAN CORPORATE LAW (1993) (strongly supporting state competition). But both sides of the debate see substantial migration and standardization. 31 efficient rules that can overcome interest group pressures might still depend on existing corporate ownership structures. The changes in legal rules that would likely induce the fiercest opposition from interest groups would be ones that directly reduce their rents. A set of rules that might be easier to pass are those that would not directly lower rents, but instead simply allow transactional changes. That is, a country may decide that instead of mandating a structure, it would allow the parties to choose their own structures. Examples might include easing rules that mandate par value, that bar certain transactions such as stock buybacks, or that ban certain ownership structures. These types of rule changes are the hardest to resist in public policy terms (because it is hard 59 to argue that having a choice is detrimental), and interest groups might be less opposed to or even favor such rules because, as long as they have sufficient control, they can ensure that rent-reducing transformations take place only if they make gains that more than offset the reduction in their rents. Some rent-reducing rules might also pass because the rent reduction is part of a larger package of legal improvements. Interest groups sometimes lose, sometimes fail to see that their ox is being gored, and sometimes are swept over in a tide of modernization. For example, a nationalist climate of self-improvement might induce political leaders to believe that the financial system must be modernized or made more competitive internationally. Because the type of financial system a country has can readily influence its corporate structures, corporate incumbents might lose if a country overhauls its financial system. Indeed, what convergence of legal rules there has been in Europe seems to fit this mode.60 Another example is that reformers may conclude that the court system must be improved across the board to facilitate commerce. Court renovation could then as a consequence protect minority stockholders (and destroy controllers’ rents) by making stockholder suits easier. Sometimes even the controllers (or the managers or the labor interests) may conclude that their lost private benefits are less than the public benefits that accrue to them with the institutional improvements. Thus, interest group obstacles to public-regarding laws are not insurmountable. But note that efficient changes might be able to overcome interest group opposition, and which form they might have to take to overcome such opposition, might still depend on the relative strength of existing interest groups—and thus in turn on the existing pattern of corporate ownership. D. Elimination of Differences in Rules by Political Fiat In the preceding Parts III.B and III.C we have shown that, in choosing legal rules, countries’ choices will depend on their existing ownership structures, and the resulting choices might consequently be path dependent and vary significantly among countries. We note in closing a qualification: that legal rules might converge if a 59 Even after such enabling laws are adopted, some efficient moves might not take place due to the interests of private parties in control. See the analysis in text accompanying notes 27-38 supra. 60[Insert citation to Draghi report and adoption of Cadbury in Italy.] 32 process of political integration leads a set of countries to agree on having an identical set of rules. That is, if lawmakers in each country are not allowed to make their own separate choices regarding corporate rules, then path dependence will disappear by political fiat. Such a process of political integration has already been taking place in Europe. While there is no question that, when countries integrate into one political system, political fiat can produce identical rules, European officials have thus far failed in 61 their efforts to end differences in corporate rules. The difficulties that European officials have encountered can be seen as a manifestation of the strength of the forces for divergence that we have analyzed. British managers, French and Italian controlling shareholders, and German codetermined firms may each prefer a system of corporate governance that radically differs from that preferred by the others. But these players might share one common position: They might wish to preserve their positional advantage in their own firms and as such might all prefer to prevent European 62 Union officials from imposing a common set of corporate rules. A simple description is instructive: The [European] Commission has been promoting the concept of the European company statute for 26 years. Successive [EU] presidents have put it on to [sic] their agendas, only to see it founder on arguments between the member states over matters such as workers’ rights. .... 61 The demise of the European Fifth Directive is discussed in J.J. Du Plessis & J. Dine, The Fate of the Draft Fifth Directive on Company Law: Accommodation Instead of Harmonisation, 1997 J. BUS. L. 23. Similarly, the proposal for a Thirteenth Directive, which was intended to unify European takeover laws, was shelved. See Proposal for a 13th Council Directive on Company Law, Concerning Takeover and Other General Bids, BULL. OF EUR. COMMUNITIES (Mar. 1989) (presented to the Council by the European Commission on January 19, 1989). The latest effort is to build a European corporate statute, with firms having the option to use the local or the EU-wide code. See John Schmid, Labor’s Equal Role Gets a Second Look in Germany: Why Keep Workers in the Boardroom?, INT’L HERALD TRIB., June 30, 1997, at 1 (discussing German reaction to the EU proposal to establish an EU-wide company statute). But see Erik Berglöf, Reforming Corporate Governance: Redirecting the European Agenda, ECON. POL’Y, Apr. 1997, at 93, 94 (“Despite recent attempts to revive the idea, hopes for Société Européenne [the European company statute] currently appear dim.”). Uniformity could come via judicial decisions that undermine the “seat of business” doctrine, which has the nation of incorporation be the nation where the firm’s principal business is located. A recent European judicial decision does open up the way to such movement for new incorporations. See Centros Ltd. v. Erhervs-og Selskabsstyrelsen [1999] E.C.R. Case C-212/97 (allowing Danish firm without British business to incorporate in Britain). If such developments take root, change might occur. 62 And countries may also become hostile to foreign structures and modes of business. French elites, for example, appear hostile to Anglo-Saxon liberalized markets and are proud of family-owned businesses that persist over generations. See Véronique Maurus, Le secret des Hénokiens, LE MONDE, Mar. 18, 1998, at 12 (noting tradition of large-firm family ownership in France). American business leaders take pride in avoiding the purportedly closed structures of continental Europe. 33 What holds up agreement is that companies do not exist in isolation but 63 are embedded in the social life of countries. In any event, regardless of how easy it is to impose identical legal rules from the center by political fiat, the analysis in this Part has focused on the common case in which lawmakers in each country are free to choose the country’s corporate rules. And in the common situation in which they are so free, their choices are likely to be path dependent. E. Conclusion on Rule-Driven Path Dependence We have shown in this Part that corporate rules, which affect choices of ownership structure, are path dependent. The choice of some corporate rules depends on the existing pattern of ownership. First, public-regarding lawmakers might often find that the existing structures, and the existing institutions that have been developed to adapt to these existing structures, affect which rules would be efficient to adopt and maintain. Second, to the extent that interest group politics affects the choice of legal rules, their dynamics and consequences might again depend on the existing ownership structures. Indeed, we have shown how this interaction between corporate ownership structures and business rules might plausibly have induced differing structures to have persisted. Thus, the two sources of path dependence of rules that we have identified can help explain why substantial differences in corporate law systems have persisted thus far. IV. OTHER BASES FOR PERSISTENT DIVERGENCE We list in this Part several other reasons for persistence of differences in corporate ownership and governance structures among the advanced economies. These reasons are not rooted in path dependence; rather, they concern ways in which some underlying parameters differ among these economies. We put them on the table for the sake of completeness and also because they reinforce the path dependence reasons for continued divergence. A. Differences of Opinion We have assumed that both lawmakers and corporate planners around the world can and could all identify which rules and which structures would be efficient. But lawmakers and corporate players genuinely disagree today, have genuinely disagreed in the past, and in all likelihood will continue to disagree as to which corporate rules and structures are best. Theory and empirical knowledge often do not tell us with confidence which corporate structure or rule would be most efficient. But without theoretical or empiri63 Stefan Wagstyl & Neil Buckley, Birthpangs of a Colossus, FIN. TIMES, July 12, 1996, at 17 (emphasis added). See also KLAUS J. HOPT, COMPANY LAW IN THE EUROPEAN UNION: HARMONIZATION OR SUBSIDIARITY (Centro di studi e ricerche di diritto comparato e staniero Conference Paper No. 31, 1998) (noting that European-wide corporate law has not arisen). 34 cal confidence, corporate players and lawmakers can genuinely disagree about which structures and rules are best. Persistent differences of opinion might well have yielded, and we suspect will probably continue to yield, persistent differences in structures and rules. Indeed, it is sufficient to look at the law review or finance literature on these subjects to see how few basic corporate issues have been resolved even 64 in the same country and culture. Now it might be argued that, even without convergence of views, natural selection might be sufficient to ensure that structures will eventually all take an efficient form. On this view, to have convergence to efficiency, players need not figure out explicitly what is optimal. Only optimal structures will survive, and natural selection will eliminate inefficient ones. People, so the argument goes, need not have understood that stores in Miami should sell swimsuits rather than furs. Stores selling the furs in Miami would have gone out of business and stores selling swimsuits would have prospered (unless there were too many of them); an equilibrium would quickly have arisen with stores selling the optimal product. But this natural selection story, although strong for stores selling furs in Miami, might not be as compelling for corporate structures and rules. Because the choice of ownership structure is only one of many aspects that will determine the success of a firm, natural selection by itself (without players recognizing the inefficiency) need not eliminate inefficient structures. Similarly, as long as players do not recognize the inefficiency of certain corporate rules, natural selection would not eliminate the economies that use these rules; such economies might become poorer on the margin, but would not be obliterated. Thus, natural selection by itself would not eliminate inefficient legal rules and ownership structures. The relatively worse performance of such rules and structures might lead to their replacement only if decisionmakers recognized that the rules and structures were indeed inefficient. And, as we discussed above, identifying which rules and structures are inefficient might be difficult not only for researchers but also for actual decisionmakers. B. Differences in Firms and Markets To focus on path-dependent reasons for divergence, we have assumed that the advanced economies on which our inquiry focuses are similar in all relevant economic conditions—and, in particular, have similar firms and markets. Dropping this assumption introduces more reasons for persistent differences. Size of economy. Some countries are smaller than others. The size of the economy influences the size distribution of its companies and the size of its capital markets. Which structure is optimal might depend on the size of a company and the 65 size of the nation’s capital markets. 64 For example, after much debate in the literature, there is still substantial difference among researchers concerning the desirable regulation of corporate takeovers. See, e.g., GILSON & BLACK, supra note 45, at 730-889. 65 Cf. Daron Acemoglu & Fabrizio Zilibotti, Was Prometheus Unbound by Chance? Risk, Diversification, and Growth, 105 J. POL. ECON. 709, 745 (1997) (offering a theory of economic development that links the degree of market incompleteness to capital accumulation and growth); William 35 What firms do. Countries might differ greatly in what their firms do and how they operate. Countries differ in their location, their natural resources and their investments in human capital. These underlying differences, as well as benefits from specialization and network externalities, might lead to differences among countries in what their corporations do. And such differences might lead to different ownership and governance structures. Optimal corporate structures and rules might depend on the type of technologies, inputs, and workforce that a company has. Thus, if countries differ systematically in their firms and technologies, then the legal rules that would be 66 most efficient for them might differ, and the corporate ownership structures that would be most efficient would differ as well. C. Differences in Culture, Ideology, and Politics We have viewed legal rules as a product of (i) public-regarding judgments as to which rules would produce the highest value, distorted by (ii) interest group politics. But we are not complete materialists. Culture and ideology, not only value maximization and self-interest, might influence a country’s choice of corporate law. American culture, for example, resists hierarchy and centralized authority more than, say, French culture. German citizens are proud of their national codetermination. Italian family firm owners may get special utility from a longstanding fam67 ily-controlled business, while an American family might prefer to cash the company earlier and run the family scion for the U.S. Senate. One link between political ideology and corporate ownership structures is ana68 lyzed by one of the authors elsewhere. According to that analysis, countries in which social democratic ideologies are dominant may empower employees more than do countries with other types of governments, putting more pressure on managers to side with employees instead of owners. As a consequence, owners may prefer their next best means of control (to resist such pressure), and that the next best means may be concentrated ownership. As such, not only might the demand for rule changes be weak in social democracies, but the demand for differing ownership and governance structures may also persist as long as the political differences persist. J. Baumol & Ralph E. Gomory, Inefficient and Locally Stable Trade Equilibria Under Scale Economies: Comparative Advantage Revisited, 49 KYKLOS 509, 510-16 (1996) (analyzing the inefficient trade equilibria produced by scale economies despite market mechanisms). 66 This statement assumes some economies of scale for corporate rules—i.e., that it would cost more to supply a separate corporate law system for each set of companies and hence each country will develop a system to best fit its typical firms. For an example of a nation’s failure to develop separate corporate law systems, see the discussion in supra note 42 (discussing the American failure to bifurcate bankruptcy into public and private firms and the eventual merger of the systems). 67. See Celestine Bohlen, A Delphic Oracle Has Seen the Future, and Likes It, N.Y. TIMES, Apr. 14, 1998, at A4 (describing how Giovanni Agnelli’s prestige is based on his family’s control of Fiat, the Italian automobile maker). 68 See Roe, Political Preconditions, supra note 20. 36 CONCLUSION We have developed a theory in this paper of path dependence of corporate ownership and governance structures. We have shown how the corporate structures that an economy has at any point in time are likely to depend on those that it had at earlier times. One type of path dependence is structure driven. We showed how an economy’s initial ownership structures directly influence subsequent choices of ownership structure. We identified two reasons for such structural path dependence—one grounded in efficiency and the other grounded in rent-seeking. First, because of sunk adaptive costs, complementarities, network externalities, endowment effects, and multiple optima, which structure is efficient depends partly on the structures with which the company and/or other companies in its environment began. Second, existing ownership structures might have persistence power, even in the face of some inefficiencies, due to internal rent-seeking. Those parties that participate in control under existing structures, as we have shown, might have an incentive and an ability to impede changes that would enhance efficiency but would reduce their private benefits of control. The other type of path dependence is rule driven. We showed that initial ownership structures affect subsequent structures also through affecting the corporate rules under which these subsequent structures will be chosen. We identified two reasons—one grounded in efficiency and the other in interest group politics—why a country’s legal rules at any point in time might be influenced by the ownership patterns that the country had at earlier times. First, even assuming that legal rules are chosen solely for efficiency reasons, the initial ownership patterns influence which corporate rules would be efficient. Second, a country’s initial pattern of corporate ownership structures influences the power that various interest groups will have in the political process that produces corporate rules. Thus, initial ownership structures that gave control to a certain group of corporate players (say, professional managers or controlling shareholders) would increase the likelihood that the country would have subsequently the rules favored by this group of players. Our analysis sheds light on why the advanced economies differ in their patterns of corporate ownership and governance. It can explain why, notwithstanding the powerful forces of globalization and efficiency, some key differences have thus far persisted. It can also provide a basis for predicting that important differences might persist in the future. Path dependence is an important force—one that students of comparative corporate governance need to recognize—in shaping corporate governance and ownership around the world. 37