Preliminary Version Comments Welcomed BROKERING, UNDERWRITING, AND INTEGRATED INVESTMENT BANKS By Chris Stefanadis* April 2004 ABSTRACT: The paper examines the role of integrated investment banks in the information production process. It is shown that integration may allow a broker and an underwriter to achieve coordination in their quality upgrading campaigns, raising their joint profits and enhancing social welfare. Furthermore, in the absence of a strict Chinese wall between underwriting and brokering, an integrated investment bank can utilize in the underwriting division any superior informational capabilities it acquires in brokering, reducing overall variable costs. A regulatory environment that does not impose a Chinese wall leads to higher profitability and social welfare. JEL Codes: G24, G28 I thank Jamie McAndrews and Steve Peristiani for helpful suggestions, and especially George Kanatas for extensive discussions of the substantive issues. I am also grateful to seminar participants at the Federal Reserve Bank of New York and Rice University. * Jones School of Management – MS 531, Rice University, 6100 Main Street, Houston, TX 77005, tel. (713) 348 6342, cstefana@rice.edu. 1. INTRODUCTION Integrated investment banks maintain both underwriting divisions, which assist firms in selling new securities issues to investors, and brokering (i.e., retail brokering and wholesale institutional sales) divisions, which assist investors in buying new securities from firms. Since the end of the stock market euphoria in 2000, integrated banks have come under increased scrutiny. In 2003, for example, New York District Attorney Eliot Spitzer, the Securities and Exchange Commission (SEC) and other regulators forced integrated investment banks to impose strict barriers between their underwriting and research departments. Critics of the Eliot Spitzer investigation, on the other hand, argue that integration can speed up information production and yield better financial research. Still, the role of integrated investment banks is not fully understood in formal economic theory. This paper develops a simple formal model to illustrate how integration may affect the informational investments of underwriters and brokers, enhancing efficiency and raising overall social welfare. To focus on the investment aspects of integration, the model assumes away any conflicts of interest within integrated banks. The various conflicts of interest, as well as the strategies available for their resolution, have been analyzed extensively in the formal literature.1 This paper, on the other hand, examines a rather overlooked issue, the impact of integration on the informational infrastructure of underwriters and brokers.2 In the game, when an entrepreneur sells a new securities issue to investors, brokers and underwriters serve as information producers for the two opposite sides of the market. Brokers receive a fee from investors to provide information on the type of the issuing entrepreneur. Underwriters, on the other hand, receive a fee from the entrepreneur to disseminate information on the entrepreneur’s type to investors, mitigating informational asymmetry. 1 Benabou and Laroque [1992] and Morgan and Stocken [2003], for example, provide a theoretical analysis of conflicts of interest. The empirical analysis of Michaely and Womack [1999] shows that research analysts from integrated investment banks with an underwriting relationship with a stock tend to be more positive in their predictions than analysts from non-affiliated brokerage houses. 2 The analysis of this paper holds even in the presence of unresolved conflicts of interest. Then, an overall assessment of integrated investment banks would entail weighing informational investment effects and conflicts of interest effects against each other. 2 The crucial ingredient of the model is the presence of a risky up-front investment in quality upgrading; the probability of success in the quality upgrading campaign of a broker or an underwriter is determined by the level of the institution’s fixed investment. A high-quality broker or underwriter has a superior informational infrastructure and is in a better position to acquire information on the type of an issuing entrepreneur than a lowquality financial institution. A high-quality informational infrastructure may reflect, for example, the presence of exceptional human resources or technological equipment. Due to the intertwined nature of the buy and the sell side, the profitability of a broker depends not only on the outcome of its own infrastructure upgrading campaign, but also on the outcome of the campaign of the underwriter. Even if a broker succeeds in upgrading, it is able to find clients for its high-quality services only when the underwriter fails. When the underwriter succeeds, on the other hand, there is no demand for the services of a high-quality broker; the underwriter neutralizes the advantage of the highquality broker’s clients by disseminating high-quality information on the issuing entrepreneur to all investors. As a result, if a broker and an underwriter are independent entities, they are plagued by a lack of coordination in their fixed investment decisions. The underwriter does not consider the negative effect that its quality upgrading investment has on the broker, choosing an excessive level of investment, while the broker chooses an inadequate level of investment. Integration, on the other hand, allows the underwriter and the broker to achieve coordination in their fixed investment campaigns, raising their profits and enhancing social welfare. Furthermore, if integration is not constrained by the presence of a Chinese wall, i.e., a strict separation between underwriting and brokering, it allows the investment bank to enjoy the benefits of a common informational infrastructure. Then, regardless of which division succeeds in upgrading, the upgraded informational capabilities are always utilized in the underwriting division. In particular, information dissemination is more efficient when it takes place through a high-quality underwriter, rather than through a high-quality broker. Because a high-quality underwriter works for the sell side, it faces a separating equilibrium with a self-selection process among entrepreneurs; only good entrepreneurs are willing to hire a 3 high-quality underwriter. This separating equilibrium entails low variable costs because the underwriter only needs to produce direct information on good entrepreneurs for the types of all entrepreneurs to be revealed. A high-quality broker, on the other hand, works for the buy side and does not encounter such a self-selection process; it needs to produce direct information on all entrepreneurs for the types of entrepreneurs to be revealed. This difference in variable costs translates into higher profits for the highquality underwriter and higher social welfare. As a result, by always utilizing its informational capabilities in the underwriting division, a fully integrated investment bank raises its profitability and enhances overall social welfare. At an empirical level, the role of integrated investment banks has been one of the most controversial issues in the U.S. financial industry since the end of the stock market euphoria in 2000.3 The discussion recently culminated in a formal investigation by New York District Attorney Eliot Spitzer, the SEC and other regulators into the research practices of integrated investment banks.4 In April 2003 investigators and investment banks reached a settlement agreement, according which, integrated investment banks will impose a strict separation between their research and underwriting departments; research analysts will report only to the brokering division (Smith, Craig and Solomon [2003]). Some critics of the integrated banking model even believe that the investigation should be expanded, forcing integrated investment banks to split up completely into independent brokers and underwriters (Augar [2002], The Economist [2002]). Unlike regulators, who tend to focus on conflicts of interest issues, integrated investment banks often defend themselves by pointing to the effects of integration on informational investment. Investment bankers note that the existing settlement agreement imposes too many restrictions on financial institutions; inhibiting integration leads to less efficient information production and lower-quality research. Underwriters, for example, may incur a cost by not being able to use the informational capabilities of the research department (The Economist [2002]). 3 The integrated investment banking model started to become prevalent in the 1980s. See Smith [2001], for example, for a description of strategic trends in the investment banking industry. 4 In the United States, the Insider Trading and Securities Fraud Enforcement Act of 1988 requires financial institutions to erect Chinese walls to prevent the misuse of material non-public information. Empirical research, however, casts doubt on the actual effectiveness of Chinese walls in limiting the flow of information within financial institutions (e.g., Seyhun [2002]). 4 This paper brings out the mechanics of the investment banker’s argument by demonstrating the impact of integration on informational investment. Furthermore, the model shows that utilizing the informational capabilities such as the know-how of research analysts of the brokering division in underwriting can be an efficient and welfare-enhancing strategy. Several integrated investment banks followed such strategies in the 1990s. In the formal literature, several articles examine the mechanics of information production when a new securities issue is sold to investors. Beatty and Ritter [1986] and Rock [1986], for example, show how underwriters may practice underpricing to mitigate the winner’s curse problem and attract uninformed bidders. Fulghieri and Spiegel [1993] and Chemmanur and Fulghieri [1994] demonstrate how underwriters may set strict standards in information production to acquire a good reputation.5 Benveniste, Busaba and Wilhelm [2002] examine the externalities that the production of information on a firm exerts on the firm’s rivals. This paper examines a different issue than the existing literature, namely the effects of integration on the information production mechanism. There are two main differences between the paper and the existing articles on information production. First, in the model, both brokers and underwriters have the opportunity to invest in information collection capabilities; the information gathering ability thus arises endogenously for both the buy and the sell side.6 In this way, the paper can illustrate the interplay between the buy and the sell side of the market with respect to their information production investments. Second, the paper uses its endogenous information production framework to examine the role of integrated investment banks, as well as the effects of regulatory restrictions, such as Chinese walls. The general idea that vertical integration can enhance coordination among firms in different stages of production is discussed in the industrial organization literature (Tirole [1988], Carlton and Perloff [2000]). In particular, when a firm makes a strategic In a different vein, Chemmanur and Fulghieri [1999] examine the optimal timing for a firm’s decision to go public. 6 In the existing literature, the buy side and the sell side do not simultaneously have information production capabilities. In Beatty and Ritter [1986], Rock [1986] and Benveniste, Busaba and Wilhelm [2002], for example, only some investors are informed ex ante; the sell side is always unable to produce information. 5 5 decision (e.g., on prices, investments, etc), it may exert an externality on firms with complementary products. Then, vertical integration may raise aggregate profits by inducing each firm to consider the effects of its strategic decisions on other firms in the supply chain. The paper draws out the implications of this general idea for the investment banking industry. There are two basic differences between this paper and the generic industrial organization literature on vertical integration. First, because a main commodity in financial transactions is information, the investment banking sector has a zero-sum aspect; when an underwriter succeeds in upgrading, it exerts a negative, rather than a positive, externality on brokers. This is different from the structure of most other industries, where quality upgrading by one firm benefits firms in other stages of the supply chain. Second, the model of this paper is tailor-made for the investment banking sector, focusing on the microstructure of financial markets. In this way, the model illustrates in detail how integration modifies the mechanics of the information production process. The paper consists of six sections. Section 2 describes the basic model, while Section 3 solves for the equilibrium of the basic model. Section 4 examines the role of integrated investment banks. Section 5 discusses an extension of the model where more than one brokers or underwriters are able to upgrade their services. Finally, Section 6 suggests some conclusions. 2. THE BASIC MODEL There are n entrepreneurs that operate in the market for one period. At the beginning of the game, each entrepreneur j (j = 1, …, n) needs 1 unit of funds to initiate an indivisible project. An entrepreneur can potentially generate income only through the project. Otherwise, if the investment is not completed, the entrepreneur does not have the capacity to earn income. The payoff of a project is either V (V > 0) or zero. In particular, when the project exhibits a high performance, it generates an income V. If, on the other hand, the project has a low performance, it yields a zero income. In Chemmanur and Fulghieri [1994], on the other hand, information production is undertaken only by the underwriter; it is the buy side that does not have the ability to produce information. 6 An entrepreneur can be of one of two types, good or bad. When a project is sponsored by a good entrepreneur, it exhibits a high performance with probability 1 – and a low performance with probability , where < 1/2. Conversely, a project that is sponsored by a bad entrepreneur exhibits a low performance with probability 1 – and a high performance with probability . It is assumed that (1 )V 1 0 V 1 , V 2. (1a) (1b) According to condition (1a), only projects sponsored by good entrepreneurs are economically viable. Condition (1b) implies that when all investors are uninformed, no projects can be financed. As we will see in Section 3, this assumption allows us to show how informational asymmetry between the entrepreneur and investors can lead to production distortions. Each entrepreneur has a small amount of private funds K at the beginning of the game. It is assumed that K is sufficiently small, so that K [(1 )V 1] , 1 2 K 1 V . 2 (2a) (2b) Condition (2a) implies that co-financing cannot be used as a signaling device. Separating equilibria in which only good entrepreneurs use K to co-finance their project, signaling their type, do not exist (see footnote 18). Condition (2b) implies that even if an entrepreneur uses K to co-finance its project, investors do not provide any funds when they are uncertain about the type of the entrepreneur (similarly to condition (1b)). From conditions (2a) and (2b), it follows that that before entering the capital market, entrepreneur j is indifferent between committing to co-finance its project and not. For 7 simplicity and without loss of generality, it is thus assumed that the entrepreneur never co-finances and only tries to obtain funds in the capital market. In particular, entrepreneur j can raise 1 unit of funds for its project from either investor j1 or investor j2. The financing investor gives 1 unit of funds to entrepreneur j. In return, the entrepreneur grants the investor a share of the project’s future income stream. Entrepreneur j chooses the investor that demands the lowest share . If the lowest share that investors demand is strictly higher than 1, the project cannot be financed and is abandoned. Otherwise, the project is financed, and after the realization of its performance, the financing investor receives a payoff V (in the case of high performance) or zero (in the case of low performance). This procedure amounts to a purchase of a new issue of entrepreneur j’s securities by the investor. An entrepreneur does not know its type at the beginning of the game. The probability that an entrepreneur is good or bad is or 1 – respectively and constitutes public information. For simplicity, I assume that = 1/2. An entrepreneur’s type is revealed, however, before the entrepreneur sells its securities in the capital market. Once an entrepreneur’s type is revealed, it is costlessly observable by the entrepreneur itself. Investors j1 and j2 are unable to observe entrepreneur j’s type directly, but may attempt to obtain this information by hiring a broker. The objective of brokers is to gather information on entrepreneur types and sell it to investors. On the sell side, after entrepreneur j discovers its type, it has the opportunity to hire an underwriter to facilitate the financing campaign. The task of an underwriter is to mitigate information asymmetry in the capital market. In particular, an underwriter tries to obtain information on entrepreneur j’s type and convey the information to all investors, i.e., to investors j1 and j2.7 An entrepreneur is unable to disseminate information to investors itself without hiring an underwriter.8 7 In practice, underwriters often purchase a new issue of securities directly from the issuing entrepreneur and then resell it to investors (firm commitment approach). The conclusions of the paper would not change if underwriters bought and resold. In particular, before underwriters buy a new issue of securities, they engage in extensive interaction with investors, trying to disseminate information on the entrepreneur and mitigate informational asymmetry in the capital market. 8 For example, entrepreneur j may lack the credibility to convey information to investors itself; it always has the incentive to convince investors that its type is good. Underwriters, on the other hand, have repeated interactions with investors because they assist numerous entrepreneurs in their financing campaigns. As a 8 I consider a dynamic financial sector where quality upgrading serves as the primary vehicle for profitability. There are a large number of brokers and underwriters in the model. For simplicity and without any loss of generality, it is assumed that at the beginning of the game, the quality of services that are offered by brokers and underwriters is zero; no broker or underwriter is able to observe entrepreneur types.9 However, a broker B and an underwriter U have the opportunity to make a fixed investment and try to develop this capability. A fixed investment could entail, for example, an attempt to develop superior research skills and analytical know-how or to create close relationships with entrepreneurs. The model is thus consistent with the empirical observation that underwriters and brokers often face up-front costs and increasing returns to scale (Rajan [1996]). For the moment, I assume that B and U are separate entities. Later on, in Section 4, I will examine the effects of integration. In practice, quality upgrading expenditures are to a large extent fixed because they have to be incurred even if the broker or the underwriter produces zero output after upgrading. When, for example, a financial institution upgrades its human capital by hiring new personnel, it may have to offer signing bonuses and employment contracts with guaranteed severance packages. Such practices imply that a financial institution will incur a significant cost even if it does not operate at all after upgrading. Other examples of up-front fixed costs include recruiting transaction costs, the costs of buying and installing new information systems, or the costs of improving the financial institution’s communication channel with investors and entrepreneurs.10 The outcome of a quality upgrading campaign is uncertain, in that a fixed investment can either succeed or fail.11 In the former case, a financial institution offers result, underwriters have a strong incentive to transmit truthful information to preserve their credibility. See, for instance, Beatty and Ritter [1986] and Chemmanur and Fulghieri [1994]. 9 Alternatively, we could assume, that initially each broker or underwriter has a probability pb or pu respectively of observing an entrepreneur’s type (0 pb < 1, 0 pu < 1). If broker B or underwriter U manages to upgrade, its probability of gaining access to information becomes higher. Such a modeling structure, however, would complicate our calculations without qualitatively affecting the conclusions. 10 Aside from entailing large fixed costs, quality upgrading may also lead to significant variable costs. This paper focuses on the fixed costs of upgrading. Chemmanur and Fulghieri [1994], on the other hand, focus on the variable costs of quality upgrading in underwriting. 11 Practitioners often point out that in the financial sector, the outcome of quality upgrading campaigns is uncertain. When a financial institution hires new personnel, for example, it cannot be certain about the exact impact of its hiring choices. Furthermore, the financial sector is often characterized by a lack of a clear strategic direction, which forces individual institutions to make their own risky decisions. The 9 high-quality services and is always able to gain access to and convey information on entrepreneur types. If the fixed investment fails, on the other hand, the financial institution finds itself no better off in terms of its prospects than it would have been had it undertaken no fixed investment whatsoever. A high-quality underwriter’s marginal cost the cost of producing information on the type of an additional entrepreneur is c > 0, where [(1 )V 1] / 2 c 0 .12 Similarly, a high-quality broker has to incur an incremental cost c to produce information on an additional entrepreneur. Once a high-quality underwriter or broker produces information on the type of an entrepreneur, it can distribute this information to any number of investors at a zero cost. Low-quality underwriters and brokers, on the other hand, are always unable to observe entrepreneur types, and their marginal cost is zero. The assumption that quality upgrading is an all-or-nothing event is made for simplicity; in the model, a high-quality financial institution is in a position to obtain information on the types of all n entrepreneurs (by incurring a marginal cost c for each entrepreneur), while a low-quality institution cannot observe the type of any entrepreneur. Our results would be exactly the same, however, if financial institutions organized separate quality upgrading campaigns for subsets of these n entrepreneurs.13 For simplicity, it is also assumed that U and B face an identical probability function in their quality upgrading efforts. The probabilities of success for U and B are p( FU ) and p( FB ) , where FU and FB are the fixed investments of U and B. The probability function is increasing in the amount of investment but at a decreasing rate, i.e., p( F ) / F 0 and 2 p( F ) / F 2 0 . The concavity of the probability function research departments of investment banks are a good example. Different investment banks are currently reorganizing their research departments in different ways. Morgan Stanley, for example, is reorganizing its research on the basis of the assumption that there will be strong demand for longer, analytical reports, rather than frequent company-specific reports (Keaveny [2003]). Other investment banks, on the other hand, are making different strategic choices. It follows that the outcome of such quality upgrading investments can be quite uncertain. 12 As we will see later, this inequality implies that in the case of success, the quality enhancement outweighs the higher marginal cost. A broker or an underwriter is thus better off if it manages to upgrade the quality of its services. 13 If the n entrepreneurs operate in different sectors, for example, a financial institution may have to engage in independent quality upgrading campaigns in each of the sectors, developing information gathering capabilities in pharmaceuticals, software, automobiles, etc. Or, in an extreme case, if the information collection synergies within a sector are not significant, a financial institution may have to organize an independent quality upgrading campaign for each individual entrepreneur. 10 ensures the existence of equilibrium in the quality upgrading stage game. U and B can choose their investments from the interval [0, F ] . Brokers and underwriters charge a fee for their services. When entrepreneur j hires an underwriter, it immediately uses its private funds to pay the underwriter’s fee. I assume that K (1 )V 1 , so that an entrepreneur always has sufficient funds to pay the underwriter’s fee. It must be stressed that the assumption that an entrepreneur pays the entire underwriter’s fee up-front is made for simplicity and without any loss of generality. In particular, it would probably be more realistic to assume that an entrepreneur pays the underwriter’s fee in two installments. First, when entrepreneur j hires an underwriter, it immediately uses its private funds to pay an amount that exactly covers the underwriter’s marginal cost. The rest of the fee is paid later by funds that entrepreneur j raises from investors j1 and j2 in the capital market (funds in addition to the 1 unit required for starting the project). This two-tier fee structure would be consistent with the standard pricing practices of actual underwriters. Such a two-tier fee schedule, however, would complicate the calculations without affecting the conclusions. So long as the two-tier structure includes an up-front fee, the conclusions are the same as if we simplify and assume that the entire underwriter’s fee is paid up-front. It is assumed that an underwriter or a broker never deceives investors by intentionally disseminating false information on an entrepreneur.14 Specifically, if a financial institution offers high-quality services, it always obtains and truthfully conveys the right information on entrepreneur types. A low-quality underwriter or broker that is unable to observe entrepreneur types, on the other hand, does not convey any information at all. For example, a financial institution may face a high non-pecuniary penalty for transmitting false information. The non-pecuniary penalty can take the form of lost reputation.15 Or, as Crockett, Harris, Mishkin and White [2003] point out, two strategies that can potentially eliminate conflicts of interest are voluntary and mandatory disclosure. 14 Several articles, on the other hand, examine the potential for conflicts of interest within financial institutions (e.g., Benabou and Laroque [1992], Michaely and Womack [1999], Morgan and Stocken [2003]). Crockett, Harris, Mishkin and White [2003] discuss several strategies for eliminating or reducing conflicts of interest. 15 The importance of reputation is paramount in underwriting and brokering. See, for example, Beatty and Ritter [1986], Fulghieri and Spiegel [1993] and Chemmanur and Fulghieri [1994]. 11 By assuming that all conflicts of interest within financial institutions have been resolved, I can focus on the main theme of the paper, the link between integration and informational investment. We have a five-stage game: Stage 1: Investment decisions by U and B. Quality upgrading campaigns take place. Stage 2: Entrepreneur types are revealed by nature. Stage 3: Entrepreneur j (j = 1, …, n) has the opportunity to hire an underwriter. Investors j1 and j2 (j = 1, …, n) have the opportunity to hire a broker. Underwriters and brokers set their fees. Stage 4: Each investor j1 and j2 specifies the share in the project’s future income stream that it demands in order to give 1 unit of funds to entrepreneur j. If the lowest share that investors demand is strictly higher than 1, the project is abandoned. Stage 5: Entrepreneur j completes the project for which he obtained funds in stage 5. Outside investors are paid according to the project’s performance. The quality of services that an underwriter or a broker offers constitutes public information. In practice, because financial institutions have repeated interactions with investors, it often becomes known whether they offer high-quality or low-quality services (Chemmanur and Fulghieri [1994]).16 Furthermore, in practice, if an underwriter or a broker succeeds in upgrading its informational infrastructure, it maintains its superior capabilities only for a certain period. After the end of this period, there can be a reshuffling of resources; resources may enter or exit the financial services sector or resources may move between financial institutions. The transfer of individual resources from a financial institution to another can be a rather lengthy and risky undertaking. The productivity of an individual resource (e.g., an employee) entails institution-specific elements and can vary significantly between different institutional environments. An individual resource exhibiting an extraordinary 16 Actually, in practice, there may be a lag between the time that a financial institution upgrades its services and the time that such an upgrading becomes public information. For example, the financial institution may need to service some entrepreneurs and engage in interactions with investors before it becomes apparent that the institution offers high-quality services. For simplicity, this paper assumes that the information lag is zero; the quality of a financial institution is public information. 12 performance in B, for example, will not necessarily exhibit the same performance if it moves to U.17 Because resources can be reshuffled, it would be more realistic to assume that we have a repeated game; after stage 5, the game goes back to stage 1 and repeats as a rearrangement of resources can take place. However, because of the chainstore paradox, the results of a finite multi-period game would be the same as those of the one-period game (see, for example, Fudenberg and Tirole [1991] for a discussion of the chainstore paradox). Thus, for simplicity and with no loss of generality, we can only consider a one-period game. I adopt the following tie-breaking conventions: (a) When an entrepreneur is indifferent between hiring a high-quality and a low-quality underwriter, or between hiring a high-quality underwriter and no underwriter at all, it hires the high-quality underwriter. (b) When an investor is indifferent between hiring a high-quality and a low-quality broker, or between hiring a high-quality broker and no broker at all, it hires the highquality broker. (c) When an underwriter or a broker is indifferent between providing its services to a client and not, it provides its services to the client. 3. EQUILIBRIUM OF THE MODEL An investor that exchanges one unit of funds for a share of a project’s future income receives a payment equal to V when the project’s performance is high. In this case, the investor’s rate of return is r = V – 1. Otherwise, when the project’s performance is low, the investor receives no payment at all. It follows that there is a one-to-one correspondence between the price that the investor charges for its funds and the rate of return r = V – 1 that the investor generates in the case that the project’s performance is high. Thus, for simplicity, we assume that when an investor gives funds to an entrepreneur, it specifies its required rate of return r = V– 1, rather than its price . The maximum rate of return an entrepreneur is able to 17 In an alternative model structure, we could assume that when B succeeds in quality upgrading, it has the opportunity to sell immediately its superior informational capabilities to U, and vice versa. As footnote 21 13 offer is rmax V 1 , which corresponds to a complete transfer of ownership to the investor, i.e., to 1. If the minimum rate of return that investors demand is strictly higher than rmax , the project cannot be financed and is abandoned. To solve for the equilibrium, we proceed by backward induction. 3.1. Competition Among Investors In stage 4, entrepreneur j (j = 1, …, n) has the opportunity to obtain 1 unit of funds by selling an ownership share to investors j1 and j2. In the price competition among investors, each investor can observe the strategy of its rival, but not the rival’s individual prices or individual rates of return. This form of competition is standard in the literature (e.g., Sharpe [1990], Von Thadden [2001]). I will first examine competition between j1 and j2 when both investors are informed. Then, I will examine competition when both investors are uninformed or when only one investor is informed. 3.1.1. Two Informed Investors Suppose that both investors j1 and j2 are informed, in that they know whether entrepreneur j is good or bad. An informed investor’s required rates of return that lead to zero expected profits are 1 rG 1 rG 1, 1 1 1 rB 1 rB 1 1, (3a) (3b) where rG and rB are the break-even required rates of return when entrepreneur j is good and bad respectively. If both investors j1 and j2 are informed and entrepreneur j is good, competition between j1 and j2 drives the equilibrium required rate of return to the break-even level rG. Entrepreneur j initiates the project, and investor profits are zero. If entrepreneur j is bad, on the other hand, the break-even required rate of return rB is higher than the maximum explains, the results of the model would be to a large extent similar. 14 rate of return an entrepreneur is able to offer, i.e., rB (1 ) / 1/ rmax V 1 (condition (1a)). A bad entrepreneur is thus not financed by investors. 3.1.2. Two Uninformed Investors Uninformed investors are unable to observe entrepreneur types. For an uninformed investor, the probability that an entrepreneur is good or bad is or 1 – respectively, where = 1/2. The break-even required rate of return for an uninformed investor is 1 r 1 r 1. 1/ 2 (4) However, r is higher than the maximum rate of return an entrepreneur is able to offer, i.e., r 1 rmax V 1 (condition (1b)). It follows that when both investors j1 and j2 are uninformed, entrepreneur j regardless of its type is never financed.18 3.1.3. One Informed Investor Suppose that only investor j1 is informed. Then, if entrepreneur j is good, j1 sets its required rate of return equal to V – 1. If entrepreneur j is bad, on the other hand, j1 does not offer funds at all. Furthermore, uninformed investor j2 does not offer any funds to entrepreneur j. It follows that a good entrepreneur j is always financed by the informed investor j1. Investor j1’s expected profit when entrepreneur j is good is (1 )V 1 0 . 18 Suppose that only good entrepreneurs co-financed their projects with private funds. Then, if a separating equilibrium existed on the basis of co-financing decisions, all investors would be informed, allowing good entrepreneurs to face a required rate of return rG in the capital market. As condition (2a) implies, however, bad entrepreneurs would have an incentive to pose as good entrepreneurs, also co-financing their projects. Such a separating equilibrium thus does not exist. Furthermore, according to condition (2b), if all investors are uninformed, no project can be financed by investors even if entrepreneurs co-finance. 15 3.2. Fee Competition in Brokering and Underwriting If both underwriter U and broker B are low-quality, we have both pooling and separating equilibria. In the pooling equilibrium, all low-quality underwriters and brokers set their fee equal to zero. Entrepreneur j chooses a low-quality underwriter or no underwriter at all at random. Then, investors j1 and j2 are uninformed, and entrepreneur j is unable to finance its project even if its type is good. Aside from the above pooling equilibrium, we can also have a class of separating equilibria. Specifically, a low-quality underwriter may charge a fee in the interval [V /(1 ), (1 )V 1] and be hired by all good entrepreneurs. Bad entrepreneurs, on the other hand, are serviced by other low-quality underwriters or by no underwriter at all at a zero fee. Bad entrepreneurs have no incentive to hire the expensive underwriter because even if they face a required rate of return rG by investors, they are unable to recover the fee. These separating equilibria are accidental and based solely on investor beliefs. Good entrepreneurs hire the expensive low-quality underwriter because investors expect them to do so, and in equilibrium, investor beliefs are realized. For simplicity, in the analysis, we will ignore such accidental separating equilibria. We will only consider separating equilibria where separation also entails actual differences between the underwriters or between the investors, rather than mere agent beliefs.19 When underwriter U is high-quality while broker B is low-quality, we have a unique separating subgame equilibrium where U is hired by all good entrepreneurs in stage 3 by setting its fee equal to Z, where Z (1 )V 1 . Low-quality underwriters and brokers set their fee equal to zero. Bad entrepreneurs hire low-quality underwriters or no underwriter at all. By hiring a high-quality underwriter in stage 3, a good entrepreneur ensures information dissemination in the capital market in stage 4. If, on the other hand, a good entrepreneur hired a low-quality underwriter or did not hire an underwriter at all in stage 3, it would be unable to receive financing for its project in stage 4.20 19 In an accidental separating equilibrium, no informational investment is necessary for entrepreneur types to be discovered and good projects to be financed. Information dissemination is completely costless. 20 This subgame equilibrium is unique. In particular, if a pooling equilibrium existed, all entrepreneurs would hire low-quality underwriters at a zero fee or no underwriter at all. U, however, would then be able to attract all good entrepreneurs by charging a fee Z. It follows that a pooling equilibrium does not exist. 16 Similarly, when both underwriter U and broker B offer high-quality services, we have the same subgame equilibrium in stage 3, where U is hired by all good entrepreneurs by setting its fee equal to Z. Investors hire either low-quality brokers by paying a zero fee or no broker at all. In particular, because we have a separating equilibrium, information on entrepreneur j’s type is revealed to the general public in stage 3; good entrepreneurs and bad entrepreneurs hire high-quality and low-quality underwriters respectively. Investors j1 and j2 are thus unwilling to pay a strictly positive fee to brokers. Broker operating profits profits without subtracting out fixed costs are zero. Suppose now that broker B is high-quality while underwriter U is low-quality. Then, in a subgame equilibrium, low-quality underwriters and brokers set their fees equal to zero, and entrepreneur j randomly chooses a low-quality underwriter or no underwriter at all. Furthermore, B sets its fee equal to Z/2, where Z (1 )V 1 . Exactly one investor either j1 or j2 hires B. If, for example, j1 is the only investor that hires B in stage 3, it earns an expected profit equal to Z/2 in stage 4; j1 earns a profit of Z and 0 when entrepreneur j is good and bad respectively. Investor j1 is thus willing to pay B a fee equal to Z/2. If, on the other hand, both investors j1 and j2 hired a high-quality broker, they would both earn a zero profit in stage 4. B is thus never hired by both investors j1 and j2. When B is high-quality while U is low-quality, we also have the class of accidental separating equilibria that we had when both B and U were low-quality. Then, B sets its fee equal to incremental cost c and is hired by no investors. B’s operating profit its profit excluding its fixed investment is zero. As we explained, however, we will ignore accidental separating equilibria in the analysis. Furthermore, we cannot have a separating equilibrium where all good entrepreneurs hire U at a fee lower than Z; U would then have the incentive to raise the fee. Finally, we cannot have a separating equilibrium where good entrepreneurs hire a low-quality underwriter. In such an equilibrium, the low-quality underwriter would charge a fee in the interval [V /(1 ),(1 )V 1] . U, however, would then be able to undercut the low-quality underwriter (as long as V /(1 ) c ) and capture the good entrepreneurs; if an entrepreneur switched from the low-quality underwriter to U, it would not be labeled as bad because U has the ability to convey the entrepreneur’s type to investors. 17 3.3. Investment Decisions In stage 1, underwriter U knows that if it upgrades its capabilities, it will be hired by all good entrepreneurs in stage 3 and will earn an expected operating profit a profit excluding its fixed investment equal to n( Z c) / 2 . Otherwise, U’s operating profit will be zero. U thus faces the following maximization problem in stage 1: Max p( FU ) n( Z c ) FU 2 {FU} p( FU ) n( Z c) 1. FU 2 (5) Broker B, on the other hand, knows that if it upgrades its capabilities, its profit will depend on whether underwriter U also succeeds in upgrading its quality. Specifically, B will earn a stage 3 operating profit a profit excluding its fixed investment equal to zero if U also succeeds, or equal to n( Z / 2 c) if U fails. B’s maximization problem in stage 1 thus is Max p( FB )[1 p( FU )]n( Z c) FB 2 {FB} p( FB ) Z [1 p( FU )]n( c) 1 . FB 2 (6) The concavity of the probability function ensures the existence and uniqueness of equilibrium in the investment game. The equilibrium levels of investment by U and B, FUN * and FBN * , can be derived by solving equations (5) and (6) simultaneously. Then, the expected profits of U and B 18 in the absence of integration are UN * p( FUN *) n( Z c ) FUN * 2 BN * p ( FBN *)[1 p ( FUN *)]n( and Z c ) FBN * 2 respectively. 4. INTEGRATED INVESTMENT BANKS In the basic model, underwriter U and broker B were two independent entities. I will now examine the effects of integration. In particular, in this section, U and B are divisions of the same integrated investment bank that engages in both underwriting and brokering activities. Integration can be either full or constrained by the presence of a Chinese wall. A fully integrated investment bank can enjoy the benefits of infrastructure sharing: when the underwriting division U succeeds in upgrading its quality, it has the opportunity to share its superior informational capabilities with the brokering division B, and vice versa. In the presence of a Chinese wall, on the other hand, although U and B are divisions of the same integrated investment bank, they are unable to share their informational infrastructure with each other. For example, under a settlement arrangement that they recently reached with New York District Attorney Eliot Spitzer, the SEC and other regulators, several major U.S. investment banks have agreed to create Chinese walls and separate their research functions from their underwriting divisions. Let us first examine equilibrium in the presence of full integration. Then, the subgame equilibria in stage 4 are the same as in the basic game. Furthermore, in stage 3, if the underwriting division U has succeeded in upgrading its quality while the brokering division B has failed, the subgame equilibrium is the same as in the basic game. The stage 3 subgame equilibrium is also the same as in the basic game if both U and B have succeeded in quality upgrading. If, however, B has succeeded while U has failed, B chooses to share its superior informational capabilities with U; U charges a fee Z and is hired by the n/2 good entrepreneurs, while B is hired by no investors. In particular, the integrated investment bank earns a higher operating profit a profit excluding its up-front fixed investment when the underwriting, rather than the brokering division, offers high-quality services (or when both divisions offer high-quality services). A high-quality underwriting division 19 generates the same overall revenues nZ / 2 as a high-quality brokering division, but its overall variable costs are lower, i.e., nc / 2 nc ; a high-quality underwriting division generates the same overall revenues by discovering the type of n/2, rather than n, entrepreneurs. At an intuitive level, in the presence of a high-quality underwriter, we have a unique separating equilibrium, where only the n/2 good entrepreneurs purchase highquality underwriting services. The type of the remaining n/2 bad entrepreneurs is revealed costlessly by the separating equilibrium without the need for information production by an underwriter. The high-quality underwriter’s operating profit is n( Z c) / 2 . When there is only a high-quality broker in the market, on the other hand, investors hire the broker to produce information on the type of all n entrepreneurs. The high-quality broker’s operating profit is n( Z / 2 c) . The maximization problem of a fully integrated investment bank in stage 1 is Max [ p( FU ) p( FB ) p( FU ) p( FB )] n( Z c ) FU FB 2 {FU , FB} p ( FU ) n( Z c ) [1 p ( FB )] 1, FU 2 (7a) p ( FB ) n( Z c ) [1 p( FU )] 1. FB 2 (7b) The equilibrium levels of investment by U and B, FUI * and FBI * , can be derived by solving equations (7a) and (7b) simultaneously. Then, the expected profit of a fully integrated investment I * [ p( FUI *) p( FBI *) p( FUI *) p( FBI *)] bank is n( Z c ) FUI * FBI * . 2 Notice that although B and U can share a common informational infrastructure, they never violate any insider trading laws in the framework of the model (see footnote 20 4). For one thing, when both divisions of the integrated investment bank have superior informational capabilities, only U has a client in equilibrium. Also, even if both B and U had a clientele, the information that B would give to its clients would be identical to the information that U would disseminate to the general public; B would thus essentially provide public, rather than proprietary, information. Let us now examine equilibrium in the presence of a Chinese wall. Then, the subgame equilibria in stages 3 and 4 are the same as in the basic game. In stage 1, U and B choose their investments to maximize the expected profit of the integrated investment bank. The maximization problem is Max p( FU ) n( Z c ) Z FU p ( FB )[1 p ( FU )]n( c) FB 2 2 {FU , FB} p( FU ) n( Z c) Z [ p( FB )n( c)] 1 , FU 2 2 (8a) p( FB ) Z [1 p( FU )]n( c) 1 . FB 2 (8b) The equilibrium levels of investment by U and B, FUCW * and FBCW * , can be derived by solving equations (8a) and (8b) simultaneously. Then, the expected profit of the integrated CW * p( FUCW *) investment bank in the presence of a Chinese wall is n( Z c ) Z FUCW * p( FBCW *)[1 p( FUCW *)]n( c) FBCW * . 2 2 In practice, as in the case of an independent B and U, if an integrated investment succeeds in upgrading the quality of its brokering or underwriting division, it maintains its superior informational capabilities only for a certain period. After the end of this period, there can be entry and exit of resources or a reshuffling of resources between divisions and financial institutions. Similarly to the case of independent brokers and underwriters, in the presence of a Chinese wall, an integrated investment bank cannot instantaneously transfer individual 21 resources from one division to another; it can only attempt transfers after a certain period. For one thing, because the performance of an individual resource may involve divisionspecific elements, the transfer of individual resources (e.g., an employee) from one division of an integrated bank to another can be a rather lengthy and risky undertaking. A resource that exhibits a high performance in the division-specific environment of brokering will not necessarily exhibit the same performance in underwriting, and vice versa. Also, the presence of a Chinese wall itself may be an additional barrier to resource mobility. Aside from preventing the brokering and underwriting division from sharing a common informational infrastructure, a Chinese wall may also limit the mobility of individual resources between the two divisions. In the absence of a Chinese wall, on the other hand, an integrated bank has a common informational infrastructure and never needs to transfer individual resources between divisions; one division can merely utilize the informational capabilities of the other.21 As before, because resources are subject to a risky rearrangement after a period, it would be more realistic to assume that we have a repeated game; after stage 5, the game returns to stage 1 and repeats. For simplicity and with no loss of generality, however, we can only consider a one-period game. Due to the chainstore paradox, the results of the one–period game are the same as those of finite multi-period game (see Fudenberg and Tirole [1991] for a discussion of the chainstore paradox). 4.1. Endogenous Integration Decisions The analysis we have just gone through assumes exogenously that U and B are either independent entities, or divisions of an integrated investment bank, without or with a Chinese wall between them. I now consider integration to be an endogenous decision of U and B. In particular, U and B will choose to integrate if the expected profits of the integrated investment bank are higher than the expected joint profits of U and B as 21 Assume that an integrated investment bank facing a Chinese wall can instantaneously and risklessly transfer individual resources between divisions. Then, the integrated bank would always transfer its superior informational capabilities from brokering to underwriting whenever B succeeded and U failed in upgrading. To be consistent, also assume that if B and U are independent entities, they can instantaneously and risklessly trade individual resources. Then, B would sell its superior informational capabilities to U at a price n(Z – c)/2 whenever B succeeded and U failed in upgrading. The results of the game are to a large extent similar in this framework. Integration allows B and U to coordinate their investment decisions, 22 independent entities. In this case, one of the financial institutions U or B will always acquire the other. Furthermore, an integrated investment bank can decide whether to create a Chinese wall between U and B depending on the effects of such a wall on profitability. We have CW * UN * BN * . To see this, notice that if U and B were independent entities and cooperatively chose their fixed investments to maximize their joint profits in stage 1, the profit maximization conditions would be exactly the same as in the case of an integrated investment bank with a Chinese wall. Because the cooperative investment decisions lead to a higher expected joint profit for U and B than the non-cooperative decisions, we have CW * UN * BN * . Proposition 1: We have CW * UN * BN * . The idea is that integration (with a Chinese wall) allows U and B to achieve better coordination in their fixed investment decisions. If U and B are independent institutions, U does not consider the negative impact its investment has on B during the quality upgrading campaign ( p( FU ) / FU p( FB )n(Z / 2 c) 0 ). As a result, U makes an excessive investment and B make an insufficient investment from a cooperative standpoint. An integrated investment bank, on the other hand, chooses the cooperative levels of fixed investment in stage 1, raising its profit. We also have I * CW * . In particular, in the absence of a Chinese wall, the integrated divisions U and B can share their informational infrastructure with each other. Then, when the brokering division B succeeds in quality upgrading while the underwriting decision U fails, the integrated investment bank earns a higher operating profit a profit without subtracting out fixed investments in the absence, rather than in the presence, of a Chinese wall, i.e., n( Z c) / 2 n( Z / 2 c) . As a result, according to the envelope theorem ( p( FB )[1 p( FU )] 0 ), we have I * CW * . raising their profits and enhancing social welfare. The difference from before is that Chinese walls have no effect on investment decisions. 23 Proposition 2: We have I * CW * . At an intuitive level, as in the Chinese wall case, full integration allows U and B to achieve coordination in their fixed investment decisions. In addition, a fully integrated investment bank enjoys the benefits of infrastructure sharing. Regardless of which division succeeds in quality upgrading, a fully integrated investment bank always utilizes its superior informational capabilities in the underwriting division, minimizing its costs and maximizing its profits.22 Corollary 1: We have I * CW * UN * BN * . When integration decisions are endogenous, U and B always choose to become a fully integrated investment bank. Proposition 3: U and B always decide to become a fully integrated investment bank. 4.2. Social Welfare So far we have shown that integration is a profitable strategy for a broker and an underwriter. I will now examine the welfare implications of government policies that allow integration in the financial industry. In particular, I will describe the effects of integration on investors, entrepreneurs and society in general. Integration has no effect on the welfare of investors j1 and j2 (j = 1, …, n). In the game, the expected profit of an investor is always equal to zero, regardless of the outcome of the upgrading campaigns in stage 1. Specifically, if U has succeeded in quality upgrading and entrepreneur j hires U in stage 3, competition among fully informed investors drives their profit to zero in stage 4. Otherwise, if j does not hire U, investors do not provide j with any funds, again earning a zero profit. Furthermore, if U has failed in quality upgrading, investor j1 or j2 earns a zero expected profit regardless of 22 In the 1990s, several integrated investment banks routinely utilized the informational resources of brokering, such as the know-how of research analysts, in underwriting. 24 whether it hires a high-quality broker by paying a fee Z/2, or a low-quality broker (or no broker at all) by paying a zero fee. Integration also has no effect on the welfare of entrepreneur j (j = 1, …, n). The expected profit of an entrepreneur is always equal to zero, regardless of the outcome of quality upgrading campaigns. In particular, a bad entrepreneur is always unable to obtain financing. If, on the other hand, entrepreneur j is good and underwriter U has succeeded in quality upgrading, j hires U in stage 3 by paying a fee Z; j’s overall expected profit is equal to zero. If U has failed in quality upgrading, a good entrepreneur j earns a zero expected profit regardless of whether there is an informed investor providing funds with a required rate of return V – 1 or two uninformed investors providing no funds in stage 4. It follows that the effects of integration on social welfare are fully captured by the profit functions of underwriter U and broker B. A social planner seeking to maximize social welfare would choose the investments that maximize the sum of profits of U and B. Notice that the model of this paper is rather simplistic and extreme, in that underwriting and brokering services differ merely with respect to their quality; there are no other dimensions of product differentiation. As a result, quality upgrading fails to benefit entrepreneurs or investors. Upgrading is profitable only for a successful underwriter or broker that captures the entire gain from success. This simple analysis, however, can illustrate in a clear and straightforward way the mechanics and social implications of integration. Let W N * , W I * and W CW * be the expected social welfare when U and B are independent institutions, a fully integrated entity or an integrated entity with a Chinese wall. Given that social welfare coincides with the sum of profits of U and B, we have W N * p( FUN *) n( Z c ) Z p( FBN *)[1 p( FUN *)]n( c) FUN * FBN * , 2 2 W I * [ p( FUI *) p( FBI *) p( FUI *) p( FBI *)] W CW * p( FUCW *) n( Z c ) FUI * FBI * , 2 (9a) (9b) n( Z c ) Z FUCW * p( FBCW *)[1 p( FUCW *)]n( c) FBCW * . (9c) 2 2 25 From corollary 1, it follows that W I * W CW * W N * . Proposition 4: We have W I * W CW * W N * . Social welfare is maximized when government policies permit the formation of fully integrated investment banks without Chinese walls. For one thing, full integration allows underwriter U and broker B to attain coordination in their investment decisions and choose the socially optimal levels of investment. Also, social welfare is maximized when investor types are revealed through a high-quality underwriter, rather than through a high-quality broker. In this way, full integration enhances social welfare by giving an integrated investment bank the flexibility to utilize in underwriting any superior informational capabilities it acquires in brokering. 5. UPGRADING BY MORE THAN ONE BROKERS OR UNDERWRITERS In the basic model, only broker B and underwriter U have the opportunity to upgrade the quality of their services. I will now examine an extension of the model in which other financial institutions may also develop the ability to gather information on entrepreneur types. As we will see, most of our conclusions are robust to the introduction of quality upgrading by other institutions. In particular, let us first suppose that there is a second underwriter U2 that has a probability q of upgrading the quality of its services. Probability q is an exogenous parameter, and underwriter U2 does not incur any fixed costs. The success or failure of U2’s campaign is determined in stage 1. As in the basic model, in stage 1, underwriter U and broker B also make a fixed investment in quality upgrading. The structure of the rest of the game is exactly the same as in the basic model. Then, the subgame equilibria in stage 4 are the same as in the basic model. In stage 3, if U2 has failed in its quality upgrading campaign, the subgame equilibria are the same as in the basic game. If U2 is the only underwriter that has upgraded its quality, on the other hand, it is hired by all good entrepreneurs at a fee Z. If both U2 and U have succeeded in upgrading, they compete in fees. Good entrepreneurs then hire either U2 or U at a fee equal to marginal cost c. 26 Let us first assume that U and B are independent entities. In stage 1, U faces the following maximization condition. Max p ( FU )(1 q) n( Z c ) FU 2 {FU} p ( FU ) n( Z c ) (1 q ) 1. FU 2 (10) The maximization condition of B is Max p( FB )(1 q )[1 p ( FU )]n( Z c) FB 2 {FB} p( FB ) Z (1 q )[1 p ( FU )]n( c) 1 . FB 2 (11) When U and B are divisions of a fully integrated investment bank, we have the following maximization condition in stage 1 Max [ p( FU ) p( FB ) p( FU ) p( FB )](1 q) n( Z c ) FU FB 2 {FU , FB} p( FU ) n( Z c ) [1 p( FB )](1 q) 1, FU 2 (12a) p( FB ) n( Z c ) [1 p( FU )](1 q) 1. FB 2 (12b) 27 When U and B are divisions of an integrated investment bank with a Chinese wall, the stage 1 maximization condition becomes Max p( FU )(1 q) n( Z c ) Z FU p( FB )[1 p( FU )](1 q)n( c) FB 2 2 {FU , FB} p( FU ) n( Z c ) Z (1 q)[ p( FB )n( c)] 1 , FU 2 2 (13a) p( FB ) Z (1 q )[1 p ( FU )]n( c) 1 . FB 2 (13b) N N Let U * and B * be the equilibrium expected profits of underwriter U and I broker B if there is no integration. Also, * and CW * are the equilibrium expected profits of a fully integrated investment bank and an integrated investment bank with a I Chinese wall. Similarly to the basic game, we have * CW N N * U * B * . As before, integration allows U and B to coordinate their investment decisions. Furthermore, in the absence of a Chinese wall, an integrated investment bank can always utilize in underwriting its informational capabilities in brokering, raising its profits. Social welfare is equal to qn( Z c) / 2 plus the sum of expected profits of U and N I B. Let W * , W * and W CW * be the expected social welfare when U and B are independent institutions, a fully integrated entity or an integrated entity with a Chinese N N N I I wall. We then have W * qn(Z c) / 2 U * B * , W * qn(Z c) / 2 * and W CW * qn(Z c) / 2 CW I * . It follows that W * W I Proposition 5: We have * CW N N CW N * W *. I * U * B * and W * W 28 CW N * W *. Let us now suppose that in the underwriting sector, only U has the opportunity to upgrade its services (as in the basic model). In the brokering sector, however, there is also a second broker B2 that has a probability q of upgrading. By following the same procedure as before, we can see that U and B can raise their joint profits by integrating. Furthermore, full integration leads to higher profits than integration with a Chinese wall. The welfare effects, however, are ambiguous because social welfare does not directly correspond to the sum of profits of U and B. Specifically, when both U and B2 offer high-quality services, U earns an operating profit a profit excluding its fixed investment equal to n( Z c) / 2 , while the social gain from U’s upgrading is only nc / 2 . This distortion leads to second-best investment decisions from a social standpoint. Because we compare second-best situations, the effects of integration on social welfare are ambiguous. Finally, let us suppose that both U2 and B2 have a probability q of upgrading. If U2 and B2 are independent entities or an integrated entity with a Chinese wall, we merely have a combination of the cases where only U2 or only B2 had a probability q of upgrading. Let us suppose, however, that U2 and B2 constitute a fully integrated investment bank. Then, as we explained in the basic model, U2 always utilizes any superior informational capabilities that B2 acquires. By following the same procedure as before, we can see that U and B also choose to become a fully integrated investment bank. Full integration leads to higher joint profits for U and B and higher social welfare than integration with a Chinese wall, which in turn leads to higher joint profits for U and B and higher social welfare than no integration. 6. CONCLUSION The paper develops a simple formal model for the role of integrated investment banks in the information production process. It is shown that integration may allow a broker and an underwriter to achieve coordination in their quality upgrading campaigns, raising their joint profits and enhancing social welfare. Moreover, in the absence of a strict Chinese wall, an integrated investment bank can utilize in underwriting any superior informational capabilities it acquires in brokering, reducing overall variable costs and further augmenting profitability and social welfare. 29 The paper focuses on the effects of integration on informational investment and does not intend to provide a general assessment of the practice. For one thing, the model assumes that integrated investment banks are in a position to resolve all internal conflicts of interest. In practice, when financial institutions are unwilling or unable to control conflicts of interest, the costs of integration may be significant. Also, the paper does not examine other types of potential synergies within integrated investment banks, such as economies of scope between underwriting and trading or the access of the underwriter to the broker’s clientele. It follows that determining the overall effects of integration is an empirical matter. Expressing, however, the link between integration and information production in formal models may lead to a better understanding and a more accurate evaluation of the integrated banking model. The debate about integrated investment banks cannot be complete unless formal models incorporate the aspects of the world that practitioners consider important. 30 REFERENCES Augar, Philip. “The City Needs a Second Big Bang,” Financial Times, October 30th, 2002, p. 21. Beatty, Randolph P., and Jay R. Ritter. “Investment Banking, Reputation, and the Underpricing of Initial Public Offerings,” Journal of Financial Economics, 15, January/February 1986, p. 213-232. Benabou, Roland, and Guy Laroque. “Using Priviledged Information to Manipulate Markets: Insiders, Gurus, and Credibility,” Quarterly Journal of Economics, 107, August 1992, p. 921-958. Benveniste, Lawrence M., Walid Y. Busaba, and William J. Wilhelm Jr. “Information Externalities and the Role of Underwriters in Primary Equity Markets,” Journal of Financial Intermediation, 11, January 2002, p. 61-86. Carlton, Dennis W., and Jeffrey M. Perloff. Modern Industrial Organization, AddisonWesley, 3rd Edition, 2000. Chemmanur, Thomas J., and Paolo Fulghieri. “Investment Bank Reputation, Information Production, and Financial Intermediation,” Journal of Finance, 49, March 1994, p. 57-79. Chemmanur, Thomas J., and Paolo Fulghieri. “A Theory of the Going-Public Decision,” Review of Financial Studies, 12, Summer 1999, p. 249-279. Crockett, Andrew, Trevor Harris, Frederic Mishkin, and Eugene White. “Conflicts of Interest in the Financial Industry: What Should We Do About Them?” Working Paper, CEPR, May 2003. Fudenberg, Drew, and Jean Tirole. Game Theory, Cambridge, MA: The MIT Press, 1991. Fulghieri, Paolo, and Matthew Spiegel. “A Theory of the Distribution of Underpriced Initial Public Offers by Investment Banks,” Journal of Economics and Management Strategy, 2, Winter 1993, p. 509-530. Keaveny, Jake. “Wall Street Research Faces Uncertain Times,” Reuters, August 27, 2003, http://biz.yahoo.com/rf/030827/bizfeature_financial_research_1.html Michaely, Roni, and Kent L. Womack. “Conflict of Interest and the Credibility of Underwriter Analyst Recommendations,” Review of Financial Studies, 12, 1999, Special Issue, p. 653-686. Morgan, John, and Phillip C. Stocken. “An Analysis of Stock Recommendations,” RAND Journal of Economics, 34, Spring 2003, p. 183-203. 31 Rajan, Raghuram G. “The Entry of Commercial Banks into the Securities Business: A Selective Survey of Theories and Evidence,” in Anthony Saunders and Ingo Walter (eds), Universal Banking: Financial System Design Reconsidered, Chicago: Irwin, 1996, p. 282-302. Rock, Kevin. “Why New Issues Are Underpriced,” Journal of Financial Economics, 15, January/February 1986, p. 187-212. Seyhun, Nejat H. “Conflicts of Interest in Securities Firms and Chinese Walls,” Working Paper, University of Michigan, School of Business, October 2002. Sharpe, Steven A. “Asymmetric Information, Bank Lending, and Implicit Contracts: A Stylized Model of Customer Relationships,” Journal of Finance, 45, September 1990, p. 1069-1087. Smith, Randall, Susanne Craig, and Deborah Solomon. “Wall Street Firms to Pay $1.4 Billion to End Inquiry,” Wall Street Journal, April 29, 2003, p. A1. Smith, Roy C. “Strategic Directions in Investment Banking: A Retrospective Analysis,” Journal of Applied Corporate Finance, 14, Spring 2001, p. 111-123. “The Price of Atonement,” The Economist, November 16th, 2002, p. 61-63. Tirole, Jean. The Theory of Industrial Organization, Cambridge, MA: MIT Press, 1988. Von Thadden, Ernst-Ludwig. “Asymmetric Information, Bank Lending, and Implicit Contracts: The Winner’s Curse,” Working Paper, Universite de Lausanne, April 2001, http://www.hec.unil.ch/evonthadden/Homepage/research/papers/sharpeaug2001.pdf. 32