WPS6350 Policy Research Working Paper 6350 Notes on Financial System Development and Political Intervention Fenghua Song Anjan Thakor The World Bank Development Economics Vice Presidency Partnerships, Capacity Building Unit January 2013 Policy Research Working Paper 6350 Abstract The paper studies the impact of political intervention Then they examine how politicians determine the nature on a financial system that consists of banks and financial of political intervention designed to expand credit markets and develops over time. In this financial system, availability. The authors find that political intervention banks and markets exhibit three forms of interaction: in banking exhibits a U-shaped pattern, where it is most they compete, they complement each other, and they co- notable in the early stage of financial system development evolve. Coevolution is generated by two new ingredients (through bank capital subsidy in exchange for state of financial system architecture relative to the existing ownership of banks) and in the advanced stage (through theories: securitization and risk-sensitive bank capital. direct lending regulation). Despite expanding credit The authors show that securitization propagates banking access, political intervention results in an increase in advances to the financial market, permitting market financial system risk and does not contribute to financial evolution to be driven by bank evolution, and market system evolution. Numerous policy implications are advances are transmitted to banks through bank capital. drawn out. This paper is a product of the Partnerships, Capacity Building Unit, Development Economics Vice Presidency. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at song@psu.edu and thakor@wustl.edu. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Notes on Financial System Development and Political Intervention∗ Fenghua Song and Anjan Thakor ∗ Fenghua Song (song@psu.edu) is Assistant Professor of Finance at Smeal College of Business, Pennsylvania State University, University Park, PA 16802. Anjan Thakor (thakor@wustl.edu; corresponding author) is Research Associate with European Corporate Governance Institute (ECGI) and John E. Simon Professor of Finance at Olin Business School, Washington University in St. Louis, Campus Box 1133, One Brookings Drive, St. Louis, MO 63130. We thank Justin Yifu Lin, Claudio Raddatz, participants at the World Bank Conference on Financial Structure and Economic Development (June 2011), and particularly two anonymous referees for their comments, and to the World Bank for �nancial support. We alone are responsible for any remaining errors. There is strong evidence that the development of the �nancial system – consisting of banks and �nancial markets – positively affects real-sector growth through the efficient mobilization and allo- cation of capital. This point was recognized at least as far back as Gurley and Shaw’s (1955, 1956) discussions of the relationship between �nancial system development and growth in the real sector. Gurley and Shaw (1955) note, “... conventional theories of income, interest, and money have given insufficient attention to important reciprocal relationships between real development and �nancial development... An additional complexity is that development of �nancial institutions, including nonbank intermediaries, is both a determined and a determining variable in the growth process.� More recently, this point has been made by various authors. For example, Gertler (1988) points out the �nancial system matters because it impinges on the resolution of informational asymmetries that have real effects, and that �nancial contracts and institutions are determined simultaneously with real variables. This view is well exempli�ed in the following statement by Levine (2010) in a debate on the role of the �nancial system: “Finance is powerful... the last few centuries demon- strate that �nancial innovation is crucial, indeed indispensable, for sustained economic growth and prosperity.� This recognition of the role of the �nancial system in economic development notwithstanding, it is less clear whether the effect of the �nancial system can be attributed to the development of capital markets or banks and other �nancial institutions, or even whether the architecture of the �nancial system (i.e., the con�guration/mix of markets and institutions within a �nancial system) matters for the real-sector growth. Complicating this picture is the fact that political intervention c in credit markets seems to be ubiquitous, and often has telling consequences (Brown and Din¸ 2005; Calomiris and Wallison 2009). This raises interesting questions about the manner in which �nancial systems evolve, how they affect economic growth, and the endogenous nature of political 1 intervention at different stages of �nancial system development. The purpose of this paper is to theoretically address these questions. To put our research in perspective, it is useful to begin by briefly reviewing the existing literature on �nancial system architecture, which deals with the mix of bank and market �nancing in the overall allocation of credit. Financial systems are typically classi�ed as bank-based or market- based, depending on the relative shares of �nancing activities performed through banks and other �nancial intermediaries as compared to those conducted through capital markets.1 An important issue in this regard has to do with which system maximizes real-sector economic growth, but on uc this the empirical evidence is mixed. For example, Beck and Levine (2002), Demirg¨ ¸-Kunt and Levine (2001), and Levine (2002) show that the positive impact of �nancial system development on economic growth is unaffected by whether the evolution of the �nancial system is due to bank or �nancial market development, and the correlation between a country’s economic growth and the architecture of its �nancial system is rather low. Deidda and Fattouh (2008) �nd, however, that a change from a bank-dominated system to one with both banks and markets can actually hurt economic growth.2 The theoretical literature on �nancial system architecture has been largely focused on examining how banks and markets compete for borrowers and the comparative capital allocational efficiencies of bank-based systems versus market-based systems, in terms of providing risk sharing (Allen and Gale 1997), generating information feedback from security prices (Boot and Thakor 1997a; Sub- rahmanyam and Titman 1999), creating �nancial innovation incentives (Boot and Thakor 1997b), resolving asset-substitution moral hazard (Boot and Thakor 1997a), and protecting borrower pro- prietary information and strengthening R&D incentives (Bhattacharya and Chiesa 1995; Yosha 1995).3 2 This “banks versus markets� distinction, based on the dominant view that banks and markets compete, clearly has powerful policy implications. It means that regulators must make choices of focus to seek the right balance between the two sectors when attempting to influence the trajectory of �nancial system development. However, this neat distinction between banks and markets, rooted in historical patterns of �nancial system evolution from many decades ago, seems to be rapidly losing relevance in an era in which securitization and widespread use of capital market �nancing by banks have largely eroded the boundaries between banks and markets, if not dissolved them completely. Mindful of these contemporary developments, we pose the following speci�c questions about �nancial system development: 1. What are the various forms of interaction that banks and markets engage in at different stages of development of the �nancial system; and 2. What forms of endogenously-determined political interventions in the �nancial system should we expect in these different stages and what are the potential consequences of such interven- tion? To address these questions, we develop a theoretical framework in which we do not restrict the interaction between banks and markets to competition, but permit them to interact in a richer manner. Our analysis reveals that banks and markets exhibit three forms of interaction. They compete, they complement each other, and they co-evolve. Strong institutions like banks are necessary for markets to perform well, and well-functioning markets are essential for banks to be induced to be sufficiently well capitalized to expand credit availability to borrowers without exposing themselves to excessive risk. Thus, the notion that one part of the �nancial system takes precedence over the other at a particular stage in terms of the development focus of policymakers is rendered obsolete. 3 Our analysis pivots on two key frictions that impede a borrower’s ability to obtain external �nancing and influence the cost of this �nancing. One is the “certi�cation friction,� which arises from the fact that there is imperfect information about borrower credit quality. As a result, a creditworthy borrower may be erroneously denied credit were it not be “certi�ed� as worthy of credit. The other is the “�nancing friction,� which arises due to uncertainty about the availability of opportunities for the borrower to enhance its project payoff. This means that good investment opportunities may be passed up, which increases a borrower’s cost of obtaining funds from the capital market. We show that banks are better at diminishing the certi�cation friction due to their credit- analysis expertise, whereas capital markets are superior at resolving the �nancing friction by pro- viding a trading venue through which informed investors’ private information about the project payoff enhancement opportunity is (noisily) impounded into security prices, thereby lowering the probability that valuable investment opportunities are passed up. Banks and markets compete to attract borrowers based on their relative advantages in resolving the two frictions. Relative to the existing theories of �nancial system architecture, the recipe for our analysis includes two new ingredients: securitization and (risk-based) bank capital. Securitization blends the roles of banks and markets by involving the bank in the origination and screening of the loan and the market in the provision of �nancing. This, in turn, generates a positive feedback loop from bank development to market development. Improvements in the bank’s screening technology, as a result of bank evolution, increase the con�dence that capital market investors have in a securitized borrower’s credit quality, which stimulates greater informed trading in the capital market. This lowers the �nancing friction and spurs capital market evolution. Bank capital connects the two sectors in a different way. As the capital market evolves, informed trading in the market increases, which reduces the �nancing friction for all seeking market �nancing, 4 including banks which now confront a lower cost of equity capital. This enables the bank to raise additional equity from the capital market to extend riskier loans that were previously eschewed, and expands the banking sector’s lending scope. Moreover, as banks lend to riskier borrowers with lower prior credit qualities, they �nd it privately optimal to improve the precision of their screening to better distinguish creditworthy borrowers from those who are not creditworthy. This stimulates evolution of the banking sector. That is, advances in the capital market are propagated to the banking sector through bank capital. The analysis thus uncovers not only competition and complementarity between bank and market �nancing, but also co-evolution. Bank evolution spurs capital market evolution, and capital market evolution in turn spurs bank evolution. That is, banks and capital markets co-evolve with each other. The limitation of the standard approach that views banks and capital markets as competitive tubs on their own bottoms in the analysis of �nancial system development becomes starkly apparent. Securitization plays a key role in the co-evolution of banks and markets in our analysis. Thus, the shadow banking system, where securitization-related activities are featured prominently, is of pivotal importance for the issues we examine. In a sense, our analysis highlights the “bright side� of the interaction between banks and markets through shadow banking. But the shadow banking system also played a key role in the recent �nancial crisis (see, for example, Stein 2010). Gennaioli, Shleifer, and Vishny (2011) present a formal model of the link between securitization and trading of asset-backed securities in the shadow banking sector and show that securitization forges market- based connections between banks, leading to a reduction in idiosyncratic risk but an increase in systemic risk. This may be viewed as the “dark side� of shadow banking. In addition, our framework also provides a setting in which we can investigate the role of political intervention in �nancial system development, something with potential rami�cations for �nancial crises and stability. This is virgin territory in the theory of �nancial system architecture. 5 Nonetheless, political intervention in the �nancial system is not only commonplace but appears to be on the rise. Credit markets are hard for politicians to resist intervening in because expanded credit availability for potential voters can generate large perceived political bene�ts. It has been empirically documented that in emerging markets, cozy ties between bankers and politicians as well as state ownership of banks allow for a great deal of political influence in credit markets (e.g., c 2005; Din¸ Brown and Din¸ c 2005; La Porta, Lopez-de-Silanes, and Shleifer 2002). And the recent �nancial crisis has shown that even advanced �nancial systems, such as the one in the U.S., are not immune to political influence. See Calomiris and Wallison (2009), for example, for a discussion of how political influences on securitization agencies like Fannie Mae and Freddie Mac, as well as on credit rating agencies and banks, gave shape to some of the initial forces that influenced the course of the 2007-09 �nancial crisis. Giving politicians a choice of how to intervene in the �nancial system, we endogenize the manner of political intervention at different stages of �nancial system development, where the political goal is to expand credit availability. We �nd that in the early stage of �nancial development, when both the certi�cation friction and the �nancing friction are large, political intervention to expand credit availability takes the form of government ownership of banks in exchange for direct capital subsidies to banks. In the intermediate stage, there is little political intervention. In the advanced stage, political intervention returns, but this time takes the form of direct lending regulation that mandates that banks invest in low-quality, high-risk (excluded) borrowers, even if doing so imposes a loss on the banking sector. The reason for this stage-dependent intervention is that the value of capital subsidies depends on the cost of equity capital for banks, which in turn depends on the stage of development of the capital market. Moreover, bank pro�tability is also dependent on the stage of �nancial market development, so the feasibility of enacting regulations that impose losses on banks is also stage dependent. In the intermediate stage, neither is bank capital sufficiently costly to make an equity 6 subsidy attractive to banks, nor are the banking sector’s pro�ts sufficiently high to absorb the losses imposed by direct lending regulations. Thus, political intervention in banking exhibits a U-shaped pattern: it is most notable in underdeveloped and highly developed �nancial systems, but it takes different forms. Differences in form notwithstanding, the effect of political intervention is the same – an increase in �nancial system risk. The rest of the paper is organized as follows. In Section I, we describe the theoretical framework. Section II analyzes the three forms of interaction between banks and markets. Section III examines political intervention at different stages of �nancial system development. Section IV concludes with a discussion of policy implications of our analysis. I. THE FRAMEWORK In this section, we describe the economic framework, with a minimum of mathematical details. The reader is referred to Song and Thakor (2010, 2011) for more formal treatments of these ideas, including proofs of results. The focus here is on explaining the logical structure that leads to these results and the economic intuition underlying them. Agents and Economic Environment Consider an economy where each borrower needs outside �nance to invest in a project. The success of the investment depends on the borrower’s type. The borrower may be either authentic or a crook, and only an authentic borrower will invest in the project, whereas a crook will just abscond with the funds, leaving nothing for the �nanciers. Each borrower’s type is its private information. The credit quality of a borrower, q ∈ [0, 1], is simply the common prior assessment about the probability with which the borrower is authentic, with a higher q representing a better credit quality. An authentic borrower will have an opportunity to enhance its project payoff, given a particular macroeconomic state; this will be discussed in more detail shortly. 7 We study a �nancial system in which there are two frictions affecting an authentic borrower’s ability to obtain external �nance and the cost of this �nancing. 1. The �rst is a certi�cation friction, which arises due to imperfect information about borrower type. As a result, even a creditworthy (i.e., authentic) borrower may be (erroneously) denied credit. 2. The second friction, �nancing friction, arises as follows. Each borrower’s project value de- pends on whether the borrower invests in costly payoff enhancement. The investment in payoff enhancement achieves its goal only if the macroeconomic condition turns out to be favorable (e.g., a strong consumer demand for the borrower’s product, and a viable supply of labor force). Absent information about the macroeconomic condition, an authentic borrower �nds it privately unpro�table to invest in payoff enhancement, since the expected return from such an investment is less than its cost, where the expectation is calculated based on the borrower’s prior assessment about the macroeconomic condition. There are traders in the capital market who �nd it personally pro�table to invest in acquiring information about the macroeconomic condition that is useful to authentic borrowers for mak- ing their payoff-enhancement investment decisions. The motivation for the informed traders is that acquisition of this information can be used by these traders to take pro�table long positions in the security issued by the borrower when the traders learn the macroeconomic condition is conducive to making the project-payoff enhancement decision. In addition to informed traders, there are also liquidity traders who are uninformed about the macroeco- nomic condition and trade for pure liquidity needs in the capital market. Trading by liquidity traders introduces noise in the price formation process and masks informed trades, but the informed traders’ information will still be (noisily) transmitted to borrowers through the equi- 8 librium prices of securities traded in the capital market. To see this more clearly, suppose the informed traders �rst observe a favorable macroeconomic condition that they conjecture the authentic borrower will exploit to enhance the project payoff, and then they take long positions in the borrower’s security that (noisily) reveal their conjecture. By observing capital market trading and the market-clearing price for its security, the authentic borrower infers the macroeconomic state and decides to make the payoff-enhancement investment, thereby rationalizing the informed traders’ initial conjecture. That is, the capital market provides borrowers valuable information through the feedback effect of security prices. The larger the number of informed traders, the greater the impact of their trading on the price of the security, so security prices become more informative about the macroeconomic condition when there are more informed traders present in the capital market. Hence, the authentic borrowers’ investment in project-payoff enhancement becomes more efficient when there are more informed traders. However, to become an informed trader requires personal investment in information acquisition by the trader, which is costly. This cost will be greater the less developed the �nancial system. As the �nancial system develops, an increasing amount of payoff-relevant information is available in the public domain for investors, so it costs less to acquire additional information. This induces more investors to become informed in more advanced �nancial systems, making borrowers’ payoff-enhancement decisions more informative. Think of the larger number of informed and sophisticated traders in the U.S. compared to say Zambia. Banks and Capital Markets Banks are better than markets at diminishing the certi�cation friction because of their expertise in credit screening that may help to attenuate asymmetric-information problems (e.g., Allen 1990; 9 Boyd and Prescott 1986; Coval and Thakor 2005; Ramakrishnan and Thakor 1984). Speci�cally, bank screening yields an informative signal that noisily reveals the borrower’s type, where the precision of the signal is increasing in the bank’s investment in a screening technology, which is pri- vately costly to the bank. Banks may also arise to reduce state-veri�cation costs that endogenously give rise to debt contracts (Townsend 1979). Nonetheless, not all information asymmetries may be eliminated by banks, so credit market distortions associated with practices like credit rationing, that do not resemble standard Walrasian market-clearing outcomes, may persist (Stiglitz and Weiss 1981). This may create a role for capital market �nancing in different forms to provide amelioration that is unattainable with bank-intermediated contracting. Capital markets are better than banks at resolving the �nancing friction by providing a setting in which security trading leads to the (noisy) impounding of informed traders’ private information about the macroeconomic condition. As explained earlier, this friction is dissipated to a greater extent in more developed �nancial systems, where borrowers learn more accurately about the macroeconomic condition and make more efficient investment decisions. We have modeled the �nancing friction resolved by the capital market purely in terms of the feedback effect of market prices as, for example, in Boot and Thakor (1997a). However, the litera- ture has highlighted various other economic functions served by markets. For example, Allen and Gale (1999) show how markets can aggregate divergent opinions about new technologies.4 This can enhance synergies between �nancial system development and real-sector development, as shown by Laeven, Levine, and Michalopoulos (2012). Such information aggregation may be particularly important when equity rather than debt is used for �nancing, since opinions may diverge more about the upside potential of a new technology. Thus, markets may rely on equity to a greater extent than banks do, even if banks were allowed to take equity positions in borrowers.5 Viewed in this light, the resolution of the �nancing friction by the market may be viewed more broadly as 10 the resolution of an information aggregation problem. Nonetheless, our speci�cation of the role of markets as communicating valuable payoff-relevant information to �rms is entirely consistent with this alternative aggregation-of-diverse-opinions viewpoint, since it is precisely this form of aggrega- tion that paves the way for the market to generate information that may not be readily available even to the �rm’s manager. Financing Sources Each borrower has three sources from which it can choose to �nance its project: (1) borrow directly from the capital market by issuing a debt security; (2) let a bank screen and certify its credit quality �rst and then borrow from the capital market via securitization; or (3) take a loan from a bank. With direct capital market �nancing, the bank is out of the picture, so there is no “bank certi�cation� provided to the borrower. With securitization, the bank screens the borrower’s credit quality �rst, and then seeks �nancing for the borrower from the capital market via securitized debt only if the screening yields a favorable outcome. The bank incurs a �xed cost to set up a trust for securitization. With a bank loan too, the bank screens the borrower’s credit quality �rst, and then extends the loan only if the screening outcome turns out to be favorable. Of course, since the screening is noisy, the bank’s credit decision is subject to both Type-1 and Type-2 errors. Thus, with both securitization and bank �nancing, the borrower’s credit quality is “certi�ed� by the bank, albeit with the recognition that the certi�cation is not perfect. A key difference between a direct bank loan and securitization is that, with the former, the bank has to raise funding for the loan, whereas with the latter, the funding comes from the capital market. For the bank loan, there is a risk- sensitive capital requirement that is increasing in borrower risk. The bank funds the rest of the loan from deposits. Deposit gathering is costly due to the cost of setting up branches, employing 11 tellers, etc, and the total cost of deposit gathering is increasing in the amount of deposits raised by the bank. As we will discuss in more depth later, bank certi�cation will influence the credit terms that the borrower can obtain from the market with securitization. Moreover, bank certi�cation also impinges on how many traders choose to become informed in the capital market and hence the accuracy of the payoff-relevant information that the borrower can infer from security prices. II. RESULTS In this section, we discuss two main results arising from our analysis of the model described in the previous section. The �rst result concerns the borrower’s choice of �nancing source. The second result characterizes the interaction between banks and capital markets: how does the evolution of one sector of the �nancial system affect the other sector, and how does a �nancial system evolve as a whole? Borrower’s Choice of Financing Source We �rst analyze a borrower’s choice of �nancing source. We focus on authentic borrowers, as crooks will make the same �nancing choices as authentic borrowers in equilibrium to avoid being identi�ed. First, consider borrowers with the lowest credit quality, i.e., borrowers in cohorts with the highest probability of having crooks. It is clear that as a borrower’s credit quality (q ) declines, a bank’s expected payoff from lending to the borrower declines as well due to the increasingly high likelihood of lending money to a crook; this is because the bank’s screening is noisy. Thus, there exists a credit quality cutoff, call it ql > 0, below which a borrower cannot obtain �nancing. This cutoff determines the banking sector’s lending scope, with a lower cutoff corresponding to a broader lending scope. 12 To examine the borrower’s choice of �nancing source beyond the credit exclusion cutoff of ql , it is useful to introduce a little notation. The bene�t to the borrower of having its loan originated by the bank is that the bank provides screening and hence “certi�es� an affirmatively-screened borrower as creditworthy, which reduces the borrower’s cost of credit. Let VS represent this screening bene�t. It is clear that ∂VS /∂q < 0, i.e., the certi�cation bene�t declines as the borrower’s observable quality improves. This screening bene�t is available only with a bank loan and securitization. Bank certi�cation produces another bene�t for the borrower if it chooses securitization. Since a certi�ed borrower is more likely to be authentic, the project-payoff enhancement conjectured by informed traders who observe a favorable macroeconomic state is more likely to be realized with a certi�ed borrower; note that a crook will not invest in payoff enhancement even if the macroeconomic condition is favorable. This means that when a borrower is bank-certi�ed and has securitized its debt, a larger number of capital market investors will acquire information about the payoff-enhancement-relevant macroeconomic condition than if the borrower �nances in the capital market without bank certi�cation. Thus, bank certi�cation enables an authentic borrower m be the to make a more informed decision about its investment in payoff enhancement. Let VP E sec be the value value of payoff enhancement to the borrower with direct market �nancing, and VP E sec (q ) > V m (q ), of payoff enhancement with securitization. Our arguments above indicate that VP E PE sec (q )/∂q > 0, and ∂V m (q )/∂q > 0. Moreover, since the borrower’s bank-certi�cation bene�t ∂VP E PE sec (q ) − V m (q )]/∂q < 0. will get smaller as q increases, we expect ∂ [VP E PE Recall that bank �nancing is also associated with a deposit-gathering cost that is increasing in the amount of deposits gathered. Denote this cost by D(q ). Since the bank faces a risk-sensitive (equity) capital requirement that is decreasing in q , it follows that ∂D(q )/∂q > 0. 13 With these preliminaries, we can write down the incremental value of bank �nancing relative to capital market �nancing as: m ∆bank = VS (q ) − D(q ) − VP E (q ), (1) and the incremental value of securitization relative to capital market �nancing as: sec m ∆sec = VS (q ) − F + VP E (q ) − VP E (q ), (2) where F is the �xed cost of setting up the securitization trust. Let us now focus on borrowers who are not subject to credit exclusion, i.e., q ≥ ql . For these borrowers, if q is relatively low, then VS (q ) is relatively high. Moreover, for low q , banks face high m (q ) will be capital requirements, so D(q ) will be relatively low. The low q also implies that VP E sec (q ) will be low too but higher than V m (q ). Hence, for low q , both ∆ relatively low; VP E PE bank and ∆sec will be strictly positive. But if F is large enough, we will have ∆bank > ∆sec . Now, as q increases, we see that m (q ) ∂ ∆bank ∂VS (q ) ∂VP E ∂D(q ) = − − < 0, (3) ∂q ∂q ∂q ∂q m (q ) sec (q ) ∂ ∆sec ∂VS (q ) ∂VP E ∂VP E = − + < 0. (4) ∂q ∂q ∂q ∂q sec (q )/∂q > 0, we see that |∂ ∆ Since ∂D(q )/∂q > 0 and ∂VP E bank /∂q | > |∂ ∆sec /∂q |. Thus, as q increases, the bene�t of bank �nancing relative to market �nancing falls faster than does the bene�t of securitization relative to market �nancing. This is intuitive. As observable borrower quality improves, banks need to keep less capital and therefore gather more deposits, which is costly and diminishes the advantage of bank �nancing relative to market �nancing. Moreover, the certi�cation bene�t of bank origination also declines as observable borrower quality goes up; this decrease in the relative advantage of bank �nancing over market �nancing also occurs with securitization. However, since the �xed cost of setting up securitization trust (F ) is invariant to 14 borrower quality, the advantage of securitization relative to market �nancing does not decline as fast as the relative advantage of bank �nancing. Figure 1 depicts these pictorially. [Figure 1 about here] Summarizing, we have our �rst result: Result 1. An authentic borrower chooses its �nancing source in equilibrium as follows: 1. There exists a low quality cutoff, ql > 0, such that an authentic borrower with q < ql cannot obtain �nancing. 2. There exists a high quality cutoff, qh > ql , such that an authentic borrower with q ≥ qh directly borrows from the capital market and the one with q ∈ [ql , qh ) approaches a bank. 3. There exists a medium quality cutoff, qm ∈ (ql , qh ), such that an authentic borrower with q ∈ [ql , qm ) prefers a bank loan and the one with q ∈ [qm , qh ) prefers securitization. In equilibrium, every crook within a prior quality cohort q chooses the same �nancing source as the authentic borrower in that cohort. Note that the equilibrium above is pooling in the sense that the observationally-indistinguishable crooks within any observable quality cohort always choose the same �nancing source as the authentic borrowers in that cohort. This is a sequential equilibrium that satis�es the universal divinity re�nement of Banks and Sobel (1987), as proved in Song and Thakor (2010). The basic idea is that the set of pricing responses that would induce defection of the crooks from the equilibrium has a larger measure than the set of responses that would induce defection of the authentic borrowers from the equilibrium. Thus, �nanciers would view a defector to be more likely to be a crook, and this means the defector would receive a worse price by defecting than staying with the equilibrium. Thus, “hiding in the crowd� is an optimal strategy for the crooks. 15 As for the �nancing-source-choice dichotomy, the intuition can be summarized as follows. Both bank screening and capital market information revelation give rise to noisy signals, and the noise- to-information ratio increases as the borrower’s observable credit quality declines. Thus, when this credit quality (q ) is sufficiently low, the borrower is deemed insufficiently creditworthy to receive credit from any source. Borrowers of very high quality go directly to the capital market. These are borrowers who may be viewed as having sufficiently high credit reputations that they do no need bank certi�cation. For example, we know that the highest-credit-quality �rms are able to avail of commercial paper as a source of direct market �nancing, but this is not available to lower-quality �rms. The intuition here is when observable credit quality is high, not only is the marginal value of bank certi�cation low, but the incentives for investors to become informed in the capital market are high, so market prices are very informative to the managers of �rms in making real decisions. This quality-cutoff division between bank and market �nancing also appears in previous research (Bolton and Freixas 2000; Boot and Thakor 1997a). However, our analysis goes a step further to argue that borrowers of intermediate quality prefer to take bank loans that are then securitized. This allows them to take advantage of bank screening and certi�cation, which stimulates information production in the capital market and hence elevates price informativeness, while at the same time avoiding bank funding which embeds the bank’s cost of complying with capital requirements. One might wonder if the result that it is the borrowers of intermediate quality who are secu- ritized is consistent with the observed securitization of highly-risky subprime mortgages preceding the recent �nancial crisis. There are two points to note on this. First, the securitization of subprime mortgages is a relatively recent phenomenon, and for much of the time period over which mortgage securitization has existed, only qualifying prime mortgages were securitized, whereas riskier mort- gages were not. Moreover, government subsidies and various other forms of political intervention played a role in the emergence of subprime-mortgage securitization, an issue that we explicitly 16 discuss later in our analysis. Second, the term “intermediate quality� is a relative term. These could still be fairly risky borrowers. The point is that there are even riskier borrowers who are served by bank loans that stay on the books of banks. Co-evolution of Banks and Markets Our previous analysis highlights competition between banks and markets. In this section, we discuss a richer form of bank-market interaction and examine collaboration and co-dependence between banks and markets, and how a �nancial system evolves as a whole. Our framework shows that markets and banks are linked in the manner of co-dependence through two channels: securitization and risk-sensitive bank capital requirements. Let us explain the intuition. With securitization, the bank certi�es a borrower’s credit quality via credit analysis and the capital market �nances the borrower, so each sector of the �nancial system operates at its best – banks focus on credit analysis and markets focus on providing �nancing. Improvements in the bank’s credit analysis technology, as a result of development in the banking sector, enhance the capital market investors’ con�dence in a securitized borrower’s credit quality. This, in turn, encourages greater informed trading in the capital market, since payoff enhancement is more prob- able with an authentic borrower, which then improves the information conveyed by the market price of securitized debt regarding the payoff-enhancement-relevant macroeconomic condition. The consequence is a lowering of the �nancing friction and a spurring of capital market evolution. That is, securitization propagates banking advances to the capital market, permitting market evolution to be driven by bank evolution. We next examine how development in the capital market affects banks. Note that banks need to raise costly equity capital from the market to support their lending, which forges a natural link between banks and markets. When the capital market develops, there is an improvement in the infrastructure for information acquisition and processing, so it costs less for investors to become 17 informed, leading to an increase in the number of informed traders. With greater informed trading, the informativeness of the price of traded claims increases, which enables the authentic borrower to extract more precise information about the payoff-enhancement opportunity. Better investment decisions by the borrower result in a lowering of the cost of equity capital for the bank lending to the borrower. This allows the bank to raise additional equity capital (that is mandated by the risk-based capital requirement) to extend riskier loans that were previously eschewed. That is, advances in the capital market enable banks to expand their lending scope (i.e., ql decreases) and lend to riskier borrowers that were previously excluded from credit availability. Moreover, as banks lend to riskier borrowers, they �nd it privately optimal to improve their credit analysis precision to distinguish more accurately between authentic borrowers and crooks, even in the absence of a general improvement in information technology. That is, development in the capital market feeds the development of the banking sector through risk-sensitive bank capital. This means that it is through (risk-sensitive) bank capital that market advances that diminish the �nancing friction end up being transmitted to banks, permitting banks to more effectively resolve the certi�cation friction for borrowers and expand their lending scope. These feedback loops generate a virtuous cycle in which each sector of the �nancial system bene�ts from the development of the other, so banks and markets not only act as competitors as Result 1 shows, but also as collaborators that complement each other. To summarize, we have: Result 2. Bank evolution spurs capital market evolution, and capital market evolution spurs bank evolution: banks and markets co-evolve with each other. The key to this co-evolution result is two-fold. On the one hand, banks need equity capital to support on-balance-sheet loans, and the capital market is the source. Advances in the capital market make this equity capital cheaper, bene�ting banks and permitting an expansion in lending scope. On the other hand, banks act as “�lters� for the capital market, screening out egregiously- 18 bad credit risks and thereby increasing the marginal return to information production for investors, leading to more efficient security prices and capital market evolution. In this sense, our analysis complements the important observation by Laeven, Levine, and Michalopoulos (2012) about the synergies between �nancial development and real-sector devel- opment. In their analysis, real-sector development makes �nancial-sector screening technologies obsolete. So pro�table investments are not �nanced, unless screening-technology innovations allow the �nancial sector to catch up. The contrast is that we focus on the interaction between banks and markets within the �nancial system, whereas they focus on the interaction between the �nancial system as a whole and the real sector. III. POLITICAL INTERVENTION In this section, we examine how political intervention can play a role at various stages of the �nancial system development. Why Do Politicians Intervene? Objective of Political Intervention Why do politicians want to intervene in the �nancial system? One reason is that politicians seek to increase their wealth and power.6 This is achieved by increasing the likelihood of getting re- elected, since a longer tenure in office allows for greater accumulation of power and influence and better opportunities for wealth accumulation both while holding office and after. In turn, this may be facilitated by increasing the number of borrowers receiving credit from the �nancial system, which then creates an incentive for politicians to intervene in the �nancial system to expand credit availability. The idea is as follows. Being able to tout expanded credit availability as a bene�t generated for potential voters provides politicians with a way to curry favor with their voter base and increase the odds of being re-elected. Even apart from such private bene�ts, politicians may be philosophically aligned with the idea that having credit available to a broader set of people is 19 a good social goal. As an example, the idea of universal homeownership for Americans has been appealing to many U.S. politicians, and this may induce them to push banks to expand credit availability to more Americans to enable a higher number of home purchases. Speci�cally, in our framework the objective of political intervention is to expand the lending scope of the �nancial system, i.e., to decrease the credit exclusion cutoff, ql : recall only borrowers with credit qualities q ∈ [ql , 1] are able to access the �nancial system for credit (either from the banking sector or the capital market), so a lower ql corresponds to a wider lending scope. Having said this, we wish to emphasize that there are plausible circumstances in which such a goal is consonant with the goal of maximizing power and wealth. How Do Politicians Intervene? What kind of intervention do politicians engage in? In terms of speci�c strategies for political intervention, we consider two sets of strategies. First, politicians may expand the lending scope of the banking sector by providing capital subsidies to banks in exchange for government ownership. The capital infusion is subsidized in the sense that the ownership taken by the government is smaller than what its equity infusion would justify in a competitive capital market. Because banks receiving subsidized equity capital have more capital to satisfy risk-based capital requirements, they are willing to lend to riskier (lower-quality) borrowers that were previously excluded. That way, politicians can influence the �nancial system to expand its lending scope. Alternatively, politicians may enact direct lending legislation that requires banks to expand lending scope and extend credit to riskier borrowers, even if doing so imposes losses on the banking sector. Banks failing to obey the regulations face revocation of their licenses to operate. As long as banks are sufficiently pro�table, the licenses will be valuable and banks will comply.7 To analyze political intervention at different stages of �nancial system development, we use the cost of acquiring information about the payoff-enhancement-relevant macroeconomic condition as 20 a measure of the stage of �nancial system development, with the cost declining as the �nancial system develops from a primitive stage to an advance stage. The idea is that this cost is high in the early stage of �nancial system development because the necessary infrastructure for obtaining information about �rms’ investment condition is still primitive and processing information can be quite costly. By contrast, in an advanced �nancial system, there is considerable information that can be acquired and processed at a relatively low cost. Political Intervention at the Early Stage of Financial System Development At the early stage of �nancial system development, the high cost of becoming an informed investor leads to a relatively small number of informed traders. With low informed trading, market prices are not very informative, so the borrower is more likely to pass up value-enhancing investment opportunities, which in turn drives up the cost of equity capital for the bank lending to the borrower. This relatively high cost of bank equity reduces the bank’s pro�t, which leads to denial of credit to high-risk (low-quality) borrowers, i.e., the bank’s lending scope is narrow at this stage (ql being relatively high). This narrow lending scope prompts politicians to intervene in the banking sector to expand its lending scope. What form of intervention will politicians �nd optimal? First, consider direct lending regulation in which politicians enact regulations that force banks to expand the lending scope without any subsidy. This strategy of intervention is unlikely to work out at the early stage, because banks’ pro�ts from lending are too low to absorb the losses such risky lending imposes on banks. If politicians force banks to lend to low-quality borrowers, the loss to the banks may lead them to voluntarily surrender their banking licenses and quit the banking industry. A second approach for politicians is to induce each bank to increase its lending scope to lend to low-quality borrowers who were initially denied credit access (i.e., those with q < ql ) by “com- pensating� the banks. We consider government ownership in exchange for subsidized equity as a 21 tool to achieve this. Speci�cally, the subsidy manifests itself through a government ownership of the bank that is smaller than what is justi�ed by the amount of equity capital subsidized, if the subsidized capital were to be raised from a competitive capital market. Note that equity capital is more costly at the early stage due to less informed trading in the �nancial market, so each bank will value subsidized equity capital relatively highly. Clearly, a larger capital subsidy corresponds to a larger increase in the banks’ lending scope, which increases the private bene�ts that politicians obtain from broader credit access. What are the consequences of an expanded bank lending scope with a government equity sub- sidy? Will it also induce banks to increase their screening precision and hence spurs capital market evolution as it does without political intervention (see Section III)? To analyze this, note that, without a government equity subsidy, a bank �nds it privately optimal to increase its screening precision as it lends to risky, previously-excluded borrowers, since the bank bears the entire cost of not being able to screen crooks accurately. With an equity subsidy, the government has part owner- ship of the bank and hence shares the cost of imprecise screening, so a bank’s incentive to improve its screening precision is weakened. As a result, we do not get the same positive propagation of development from banks to capital markets, and the �nancial system does not experience the same evolutionary boost, despite its lending scope being expanded. In fact, the larger the government equity subsidy, the less likely it is that the banking sector will improve its screening technology. Thus, political intervention at the early stage of a �nancial system may retard its development, despite a broader lending scope. Moreover, since the expansion of the banking sector’s lending scope is not accompanied by a concomitant increase in bank screening precision, systemic risk in the �nancial system increases as well. Political Intervention at the Intermediate Stage of Financial System Development 22 The cost of becoming informed is lower at the intermediate stage, so there are more informed traders in the capital market than in the early stage of development. Thus, equity capital is less expensive for banks in the intermediate stage than in the early stage. As our previous analysis shows, the �nancial system’s lending scope is broader (lower ql ) in the intermediate stage. Politicians, however, may want to expand the scope even further. But an equity subsidy is now less attractive to a bank, and for each unit of equity subsidized, the bank responds by increasing its lending scope less in the intermediate stage than in the early stage. An equity subsidy is thus less efficient for politicians in the intermediate stage than in the early stage. Direct regulation may not work either because banks are not yet pro�table enough to absorb the accompanying losses. Thus, at the intermediate stage, we see that political intervention is muted compared to the early stage. It is even possible that when equity capital is sufficiently less expensive for banks, an equity subsidy from the government may disappear altogether at the intermediate stage. At the same time, if banks’ pro�ts are still not high enough to absorb the loss associated with direct lending regulation, direct lending regulation that mandates lending to high-risk borrowers will not be feasible either. That is, there may be no political intervention at all at some speci�c intermediate stage of �nancial system development. Political Intervention at the Advanced Stage of Financial System Development At an advanced stage of �nancial system development, information acquisition in the capital market is the least costly, so the number of informed traders in the capital market is the largest. Thus, politicians view a capital subsidy as even less attractive now than at the intermediate stage. The relatively low cost of bank equity generates relatively high pro�ts for banks in this stage of �nancial system development. Direct lending regulations that mandate that banks lend to risky, excluded borrowers now become feasible. Banks comply because their high pro�ts make their banking licenses valuable. 23 What are the consequences of the expansion of the banking sector’s lending scope at this stage of �nancial system development? Note that although banks bear the full cost of imprecise screening as they lend to risky, previously-excluded borrowers under the lending regulation, they will not increase their screening precision by investing more in the screening technology. To see why, note that if banks were to improve their screening precision when they are compelled to expand their lending scope, they would have done so voluntarily even without such regulatory pressure. The reason why banks had chosen not to lend to those highly risky (previously-excluded) borrowers in the �rst place is because the marginal bene�t of improved screening is outweighed by its marginal cost. Direct lending regulations compel banks to lend but do not fundamentally alter this bene�t- cost tradeoff. As a result, political invention at the advanced stage of �nancial system development also increases �nancial system risk, as it does at the early stage. We can summarize this discussion as follow: Result 3. Political intervention in the �nancial system is stage dependent: 1. At an early stage of �nancial system development, when information acquisition and process- ing in the capital market are sufficiently costly so informed trading is sufficiently low and bank equity cost is sufficiently high, politicians intervene by providing subsidized capital to banks. 2. At an intermediate stage of �nancial system development, when information acquisition and processing cost in the capital market increases but banks’ pro�t is still low, politicians avoid intervening with either equity subsidy or direct lending regulations. 3. At an advanced stage of �nancial system development, when information acquisition and processing cost becomes even lower and banks’ pro�ts become higher, politicians intervene by enacting direct lending regulations to force banks to expand lending scope to low-quality borrowers that create losses for banks. 24 To capture the intuition for this result in a summary form, note that the choice of political intervention rests on the tradeoff between the cost to the taxpayers/voters of the chosen intervention and the effectiveness of the intervention in influencing banks to expand lending scope. At an early stage of �nancial development, bank lending scope is constrained by the high cost of raising the equity capital needed to support risky lending. Because capital is a costly resource, politicians are able to signi�cantly influence banks by providing it at a subsidy via government ownership of banks. So that is the chosen mode of intervention. At an advanced stage, capital is cheap, so offering that as a carrot to banks to expand lending scope is far less effective. But banks are also more pro�table at this stage, so a license to operate has high value, and politicians can simply enact regulations requiring banks to engage in risky lending (even at an expected loss) because there is a sufficient pro�t cushion for each bank to satisfy its participation constraint. It is in the middle – the intermediate stage of development – that capital is not costly enough and bank pro�ts are not high enough to make either intervention strategy viable. IV. POLICY IMPLICATIONS AND CONCLUSION In this paper we have discussed the theoretical underpinnings of �nancial system evolution. The main result is that banks and capital markets within the �nancial system exhibit three forms of interaction: competition, complementarity and co-evolution. This is in contrast to the standard view that banks and markets only compete with each other. This new perspective is rich in implications about development policy, but as our analysis indi- cates, we should be careful to distinguish between policy interventions aimed at correcting market failures and those driven by political economy factors. In the past, development agencies have sometimes adopted the view that emerging capitalistic economies, like say Romania and other for- mer Soviet-bloc countries, needed to develop capital markets �rst in order to have the �nancial systems needed to support vigorous economic growth (e.g., see Meyendorff and Thakor 2002). In 25 many instances, this led to considerable resources being devoted to developing stock exchanges and over-the-counter markets like RASDAQ, the Romanian version of NASDAQ, well before robust �nancial institutions were developed. What our analysis shows, consistent with the policy conclu- sions in Meyendorff and Thakor (2002), is that such a policy prescription is misguided. Banks and markets have to be developed together; focusing �rst on markets and not on banks actually runs the risk of retarding capital market growth itself. With a policy not focused on co-evolution, one may sacri�ce the efficacy of both market and banking system development. Thus, in the interest of correcting potential market failure, development policy should focus on the simultaneous develop- ment of banks and markets. What is interesting is that this can also deal effectively with political economy forces, since it can make political intervention less likely as politicians would be unable to use the perceived failure of the �nancial system as a politically convenient reason to intervene. This brings us to our examination of endogenous political intervention in �nancial system de- velopment. Such intervention can be broadly viewed as being one of two types – either motivated by the desire to correct a market failure or motivated by the sel�sh interests of politicians. An example of the former motivation is the liquidity-provision intervention by the Federal Reserve during the recent crisis when the private provision of liquidity in the shadow banking system dried up. Such intervention is ex post efficient in the sense that it minimizes market-failure-related dis- tortions. This is not the kind of political intervention that our analysis focuses on. Speci�cally, the intervention in our analysis is motivated by the sel�sh interests of politicians – the desire to (ineffi- ciently) expand credit availability to enhance the re-electability prospects of politicians. We showed that political intervention is just as likely in well-developed �nancial systems as it is in emerging, underdeveloped systems. In emerging �nancial systems, politicians provide subsidized equity in exchange for ownership of banks. In well-developed �nancial system, politicians enact regulations that mandate that banks lend to previously-excluded high-risk borrowers. In both cases, the effect 26 is the same – systemic risk of the �nancial system is increased. Thus, if the goal is to improve �nancial system stability, the discussion cannot proceed without explicitly considering political in- centives to undertake initiatives that increase �nancial system risk. In case this increased systemic risk leads to a �nancial crisis, there may be an opportunity for political intervention that is aimed at correcting a market failure. There can thus be an interaction between the two motivations for policy intervention by the government. Purely politically-motivated intervention can increase the likelihood of intervention aimed at correcting market failures. Also interesting is the potential interaction between policy mis-steps due to the failure to recognize that banks and markets must co-evolve on the one hand and the likelihood and nature of political intervention in the �nancial system on the other hand, an issue touched upon briefly earlier. This takes us beyond the issues analyzed earlier because the nature of political intervention now becomes less predictable, since it is likely to be targeted at whatever is perceived to be a failure in the �nancial system. This introduces a form of endogenous uncertainty into the �nancial system and further elevates systemic risk. A key point emerging from our analysis is that securitization is one of the elements that is responsible for the virtuous loop that positively connects the co-evolution of banks and markets. Thus, if there is any government intervention that our analysis supports, it is in the form of encour- aging the development of securitization markets, and not any direct intervention to expand bank lending either through government ownership of banks or direct lending legislation that compels banks to lend to otherwise-excluded high-risk borrowers.8 The reason is that the encouragement of securitization helps to improve economic efficiency by reducing the impact of frictions related to asymmetric information that can lead to market failures (e.g., credit rationing or other forms of credit exclusion due to a limited scope of direct bank lending), whereas direct intervention to expand credit availability is motivated solely by political considerations and comes at the expense 27 of economic efficiency. In other words, politicians need to distinguish between market failures arising from coordination failures among pro�t-maximizing participants – that may be subject to attenuation through selective ex post government intervention – and those that are linked to the endogenous uncertainty created by government mis-steps. We have not considered the impact of the judicial system on the interactions we have examined, so a few remarks on that are appropriate. As La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998) have shown, legal systems and the implied property rights have important consequences for �nance and growth. There are at least two major issues here. One is the impact of the judicial system on how much information about potential borrowers is available in the public domain. Laws differ across countries when it comes to how much credit information lenders, credit reporting bureaus and others can share about borrowers. Our analysis suggests that the more the judicial system permits the release and dissemination of such information, the smaller will be the (screening) advantage of banks over markets. Hence, capital markets will experience a greater share of total credit extension in economies with judicial systems of this sort. The other issue is creditor versus borrower (shareholder) rights. Again, legal systems differ in the extent to which they favor one group over the other (La Porta, Lopez-de-Silanes, Shleifer, and Vishny 1998). Since the economic forces that propel banks and markets to develop further originate from creditors in our model, the analysis hints at the superiority of legal systems that favor creditor rights, at least in terms of correcting market failures and facilitating bank-market co-evolution. However, this is no more than a hint, so it would be an interesting one to examine in future research. 28 Notes 1 Typical examples of market-based systems are those in U.S. and U.K., and Japan and Eurozone countries like Germany are traditionally viewed as having bank-based systems. 2 See also the growing body of other empirical research on �nancial system architecture as well as its impact on growth (Levine and Zervos 1998; Tadesse 2002). 3 See also Allen and Gale (1999), Dewatripont and Maskin (1995), and von Thadden (1995). 4 Other papers have also used heterogeneous prior beliefs to explore issues in banking, e.g., Song and Thakor (2007). 5 Since credit rationing in models like Stiglitz and Weiss (1981) results from the use of debt contracts, the market can eliminate it by relying on equity. 6 A direct, albeit crude, way to do this would be to accept bribes from certain groups and engage in other illegal activities. While this is a realistic alternative speci�cation, we rule it out in our analysis. 7 Clearly, the intervention strategy space of politicians is much richer in practice than the two strategies we have chosen to analyze. However, we believe those are two commonly deployed strategies, and analyzing a more complex set of strategies becomes analytically intractable. 8 Clearly, our model does not argue that ex post efficient government interventions during �nancial crises are not worthwhile. 29 REFERENCES Allen, Franklin. 1990. “The Market for Information and the Origin of Financial Intermediation.� Journal of Financial Intermediation 1(1): 3-30. Allen, Franklin, and Douglas Gale. 1997. “Financial Markets, Intermediaries, and Intertemporal Smooth- ing.� Journal of Political Economy 105(3): 523-46. Allen, Franklin, and Douglas Gale. 1999. “Diversity of Opinion and Financing of New Technologies.� Journal of Financial Intermediation 8(1-2): 68-89. Banks, Jeffrey, and Joel Sobel. 1987. “Equilibrium Selection in Signaling Games.� Econometrica 55(3): 647-61. Beck, Thorsten, and Ross Levine. 2002. “Industry Growth and Capital Allocation: Does Having a Market- or Bank-Based System Matter?� Journal of Financial Economics 64(2): 147-80. Bhattacharya, Sudipto, and Gabriella Chiesa. 1995. “Proprietary Information, Financial Intermediation, and Research Incentives.� Journal of Financial Intermediation 4(4): 328-57. Bolton, Patrick, and Xavier Freixas. 2000. “Equity, Bonds, and Bank Debt: Capital Structure and Financial Market Equilibrium under Asymmetric Information.� Journal of Political Economy 108(2): 324-51. Boot, Arnoud, and Anjan Thakor. 1997a. “Financial System Architecture.� Review of Financial Studies 10(3): 693-733. Boot, Arnoud, and Anjan Thakor. 1997b. “Banking Scope and Financial Innovation.� Review of Financial Studies 10(4): 1099-131. Boyd, John, and Edward Prescott. 1986. “Financial Intermediary-Coalitions.� Journal of Economic Theory 38(2): 211-32. c. 2005. “The Politics of Bank Failures: Evidence from Emerging Markets.� Brown, Craig, and Serdar Din¸ Quarterly Journal of Economics 120(4): 1413-44. Calomiris, Charles, and Peter Wallison. 2009. “The Last Trillion-Dollar Commitment: The Destruction of Fannie Mae and Freddie Mac.� Journal of Structured Finance 15(1): 71-80. Coval, Joshua, and Anjan Thakor. 2005. “Financial Intermediation as a Beliefs-Bridge between Optimists and Pessimists.� Journal of Financial Economics 75(3): 535-69. 30 Deidda, Luca, and Bassam Fattouh. 2008. “Banks, Financial Markets and Growth.� Journal of Financial Intermediation 17(1): 6-36. u¸ Demirg¨ c-Kunt, Asli, and Ross Levine. 2001. Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. Cambridge, MA: The MIT Press. Dewatripont, Mathias, and Eric Maskin. 1995. “Credit and Efficiency in Centralized and Decentralized Economies.� Review of Economic Studies 62(4): 541-55. c, Serdar. 2005. “Politicians and Banks: Political Influences on Government-Owned Banks in Emerging Din¸ Markets.� Journal of Financial Economics 77(2): 453-79. Gennaioli, Nicola, Andrei Shleifer, and Robert Vishny. 2011. “A Model of Shadow Banking.� Working Paper, Harvard University, Cambridge, MA. Gertler, Mark. 1988. “Financial Structure and Aggregate Economic Activity: An Overview.� Journal of Money, Credit and Banking 20(3): 559-88. Gurley, John, and Edward Shaw. 1955. “Financial Aspects of Economic Development.� American Economic Review 45(4): 515-38. Gurley, John, and Edward Shaw. 1956. “Financial Intermediaries and the Saving-Investment Process.� Journal of Finance 11(2): 257-76. La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2002. “Government Ownership of Banks.� Journal of Finance 57(1): 265-301. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny. 1998. “Law and Finance.� Journal of Political Economy 106(6): 1113-55. Laeven, Luc, Ross Levine, and Stelios Michalopoulos. 2012. “Financial Innovation and Endogenous Growth.� Working Paper, Brown University, Providence, RI. Levine, Ross. 2002. “Bank-Based or Market-Based Financial Systems: Which Is Better?� Journal of Financial Intermediation 11(4): 398-428. Levine, Ross. 2010. “Financial Innovation Boosts Economic Growth: A Debate with Joseph Stiglitz.� The Economist (February – March). London: The Economist Newspaper Ltd. 31 Levine, Ross, and Sara Zervos. 1998. “Stock Markets, Banks, and Economic Growth.� American Economic Review 88(3): 537-58. Meyendorff, Anna, and Anjan Thakor. 2002. Designing Financial Systems in Transition Economies: Strate- gies for Reform in Central and Eastern Europe. Cambridge, MA: The MIT Press. Ramakrishnan, Ram, and Anjan Thakor. 1984. “Information Reliability and a Theory of Financial Inter- mediation.� Review of Economic Studies 51(3): 415-32. Song, Fenghua, and Anjan Thakor. 2007. “Relationship Banking, Fragility, and the Asset-Liability Matching Problem.� Review of Financial Studies 20(6): 2129-77. Song, Fenghua, and Anjan Thakor. 2010. “Financial System Architecture and the Co-evolution of Banks and Capital Markets.� The Economic Journal 120(547): 1021-55. Song, Fenghua, and Anjan Thakor. 2011. “Financial Markets, Banks, and Politicians.� Working Paper, Pennsylvania State University, University Park, PA. Stein, Jeremy. 2010. “Securitization, Shadow Banking, and Financial Fragility.� Daedalus 139(4): 41-51. Stiglitz, Joseph, and Andrew Weiss. 1981. “Credit Rationing in Markets with Imperfect Information.� American Economic Review 71(3): 393-410. Subrahmanyam, Avanidhar, and Sheridan Titman. 1999. “The Going-Public Decision and the Development of Financial Markets.� Journal of Finance 54(3): 1045-82. Tadesse, Solomon. 2002. “Financial Architecture and Economic Performance: International Evidence.� Journal of Financial Intermediation 11(4): 429-54. Townsend, Robert. 1979. “Optimal Contracts and Competitive Markets with Costly State Veri�cation.� Journal of Economic Theory 21(2): 265-93. von Thadden, Ernst-Ludwig. 1995. “Long-Term Contracts, Short-Term Investment and Monitoring.� Re- view of Economic Studies 62(4): 557-75. Yosha, Oved. 1995. “Information Disclosure Costs and the Choice of Financing Source.� Journal of Financial Intermediation 4(1): 3-20. 32 FIGURE 1. Borrower's Choice of Financing Source Benefit of financing source relative to market financing C Benefit of bank financing relative to market financing (CD) A Benefit of securitization relative + to market financing (AB) 0 D B − Borrower gets Borrower uses Borrower obtains direct bank loan securitization market financing 𝑞 = 𝑞𝑙 𝑞 = 𝑞𝑚 𝑞 = 𝑞ℎ Observable borrower quality 𝑞