Bank-based versus Market-based Financial Systems: A Growth-theoretic Analysisasteriskmath Shscascnsckschscasc Chscasckscrscascbscoscrsctscysc† Trsciscdsciscpsc Rascysc‡ Abstract We study bank-based and market-based financial systems in an endogenous growth model. Lending to firms is fraught with moral hazard as owner-managers may re- duce investment profitability to enjoy private benefits. Bank monitoring partially re- solves the agency problem, while market-finance is more ‘hands-off’. A bank-based or market-based system emerges from firm-financing choices. It is not possible to say unequivocally which of the two systems is better for growth. The growth rate depends, crucially, on the efficiency of financial and legal institutions. But a bank-based system outperforms a market-based one along other dimensions. Investment and per capita in- come are higher, and income inequality lower, under a bank-based system. Bank-based systems are more conducive for broad-based industrialization. A temporary income re- distribution, under both financial systems, results in permanent improvement in per capita income as well as income distribution. Kescyscwscoscrscdscssc: Financial System, Income Distribution, Banks, Market Finance JEL Clscascsscssciscfscisccscasctsciscoscnsc: E22, G20, O15, O16, asteriskmathWe thank Leonard Cheng, Jangok Cho, David Cook, Sudipto Dasgupta, George Evans, Vidhan Goyal, Jo Anna Gray, Philippe Marcoul, Nobuhiro Kiyotaki, Francis Lui, Cheol Park, Xiaodong Zhu and seminar participants at the Delhi School of Economics, Hong Kong University of Science and Technology, Iowa State, Indian Statistical Institute (Delhi), University of Hong Kong, the 2001 NEUDC Conference (Boston) and the 2002 Midwest Macro Conference (Vanderbilt), for helpful comments and discussions. The usual caveat applies. †Department of Economics, University of Oregon, Eugene, OR 97403-1285. shankhac@oregon.uoregon.edu ‡Department of Economics, Hong Kong University of Science & Technology, Clear Water Bay, Kowloon, Hong Kong. tridip@ust.hk 1. Introduction Thispaperisatheoreticalanalysisofbank-based and market-based financial systems in economic growth and development. We are motivated to study this particular issue because of a long-standing debate on the relative importance of the two systems. The success of market-based systems in the US and UK have led some observers to tout their virtues, while others have advocated bank- based systems because of their vital role in German and Japanese industrialization.1 Eastern Europe and Latin America’s financial liberalization of the 1990s has revived this debate — market-based systems are being seen as more dependable for growth and development (Allen and Gale, 2000). We examine this debate in an endogenous growth model. Availability of bank and/or market finance enables firms to make larger investment than is possible otherwise. It also allows entrepreneurs in the traditional sector to overcome large-scale investment requirements of entering the modern sector. A bank-based or market-based system emerges endogenously. Two countries may have different financial regimes, yet enjoy similar rates of economic progress; what matters for growth is the efficiency with which a country’s financial and legal institutions resolve agency problems, rather than the type of system it relies upon. From a development perspective, however, a bank-based system outperforms a market-based one — financial intermediation creates an environment more conducive for transforming a traditional economy into a modern one. With the availability of systematic evidence during the past decade, the relevance of fi- nance for development is now widely accepted. This evidence shows a positive and robust relationship between financial development and growth, the level of financial development being a good predictor of future growth and technological change.2 Concurrently, an ex- tensive theoretical literature on financial institutions has developed. Research in corporate finance has examined firm financing choices, while growth theorists have studied the role of finance in capital and knowledge accumulation. In corporate finance, the organization of financial activities is seen to affect growth through corporate governance and a firm’s ability to raise external funds. Financial interme- diaries reduce costs of acquiring and processing information about firms and their managers 1See Allen and Gale (2000), Holmstrom (1996) and Levine (2000) for details. 2Gerschenkron (1962), Goldsmith (1969) and Gurley and Shaw (1955) were among the first to discuss the importance of finance for growth; a comprehensive summary of more recent findings is contained in Levine (1997). 1 and thereby reduce agency costs (Boyd and Prescott, 1986; Diamond, 1984). By assuming the role of a ‘delegated monitor’, they also avoid wasteful duplication of information. In Holmstrom and Tirole (1997), bank- and market-finance are distinguished according to their information content. Bank monitoring resolves moral-hazard problems at the level of the firm. Firms with lower marketable collateral and higher incentive problems borrow from banks, while wealthier firms rely on unintermediated market-finance. Hence, as Boot and Thakor (1997) point out, bank lending is likely to be important when investors face ex post moral hazard problems, with firms of higher observable qualities borrowing from the capital market. Some authors have also highlighted how market finance creates appropriate incentives for a firm. In Scharfstein (1988), equity markets encourage corporate governance through hostile takeovers of under-performing firms. Rajan and Zingales (1998b, 1999) argue that market-finance transmits price signals which guides firms into making worthwhile invest- ments. Relationship-based bank finance, in contrast, could lead firms facing weak cash flows to undertake misguided investments. Contributions on finance and growth are many.3 Among them, Greenwood and Jovanovic (1990), Bencivenga and Smith (1991) and de la Fuente and Marin (1996) show that financial intermediaries promote growth by pooling risks, providing liquidity and monitoring risky innovations. Greenwood and Smith (1997), on the other hand, analyze how financial markets assist growth through increased specialization. But growth theory has been largely silent on the ‘bank versus market’ debate, stressing the importance of either banks or financial markets.4 In recent years, policymakers have been advocating a shift toward financial markets, particularly in Latin America and Eastern Europe where financial systems similar to those in the US have been proposed (Allen and Gale, 2000). It is, however, unclear why market- based systems necessarily dominate bank-based ones. As Levine (1997, pp. 702-703) points out, “we do not have adequate theories of why different financial structures emerge or why financial structures change...we need models that elucidate the conditions, if any, under which different financial structures are better at mitigating information and transaction costs.” It is precisely here that the chief contribution of our paper lies. 3This literature also branches into analyzing the role of credit markets in business cycles. See Williamson (1986, 1987), Bernanke and Gertler (1989) for early contributions and Bernanke et al. (2000) for a recent overview. 4Boyd and Smith (1998) do allow for a simultaneous choice of bank- and market-finance. But they focus on how the mix changes over time. 2 Finance is relevant for growth and development for two main reasons. Better developed financial systems resolve agency problems better. This enables firms to borrow at cheaper rates and invest more.5 But finance also plays a role in structural transformation. A key characteristic of developing countries is their dualistic structure — the coexistence of pockets of modern manufacturing activity with widespread peasant farming and cottage-industry production. Transition from the traditional sector to the modern is not always simple: manufacturing activities usually require lumpy investments that may not be forthcoming in the absence of well-developed financial systems.6 These complementary roles are built into an Ak-type endogenous growth model with overlapping generations of families. A set of agents (entrepreneurs) convert final goods into capital using a modern-sector technology that requires a minimum investment size to cover setup costs. Entrepreneurs who are able to obtain the requisite amount of finance enter the modern-sector and produce capital. Those who do not, engage in traditional activities like peasant farming and household production. As in the corporate finance literature, we distinguish between bank-finance and market- finance based upon their involvement with investment projects. Banks are typically more ‘hands-on’, engaged in project selection, monitoring firms and identifying promising en- trepreneurs. On the other hand, investment through the purchase of tradeable securities, or market-finance, is more of an arm’s length transaction, with very little subsequent in- volvement in a firm’s investment decisions. To justify a role for both types of finance, we introduce moral hazard at the level of the entrepreneur. Specifically, we adapt Holmstrom and Tirole’s (1997) agency problem to a dynamic setting. Entrepreneurs who obtain external finance may deliberately reduce the success probability of their investment in order to enjoy private benefits. Outside investors, or the market, are too disparate to effectively control a borrower’s activities. Financial intermediaries, on the other hand, monitor entrepreneurs and (partially) resolve the agency problem. But since monitoring is costly, bank finance is more expensive than market finance. A key determinant of financing choices is an entrepreneur’s initial wealth. Entrepreneurs with lower wealth have more incentive to be self-serving than wealthier ones. One way to 5Rajan and Zingales (1998a) use industry-level data to show that more developed financial regimes pro- mote growth by reducing the cost of borrowing. 6The effect of capital market imperfections on inequality have been studied by Galor and Zeira (1993) for human capital investment, and Banerjee and Newman (1993) for occupational choice. Hicks (1969) as well as North (1981) deliberate on the role of finance in overcoming large-scale investment requirements during the Industrial Revolution. 3 mitigate this incentive gap is to borrow, at a higher rate, from a bank and agree to being monitored. In contrast, wealthier entrepreneurs rely more on market finance as they face less of an information gap. In certain cases, for instance when the fixed cost of modern sector activities are large, even bank monitoring is not a sufficient substitute for entrepreneurial wealth — the poorest entrepreneurs are unable to get any type of external finance.7 A bank-based system, where intermediation plays a key role, or a market-based system, where all lending is unintermediated, evolve endogenously in our model. A bank-based sys- tem emerges when monitoring costs are modest and when agency problems are significantly extenuated through monitoring. When agency problems are not particularly severe, or when monitoring is expensive, a market-based system emerges. The growth rate under either regime is a function of the efficiency of the system — better- functioning legal systems make contracts easier to enforce and reduce monitoring costs as also the cost of direct lending. Investment is higher, as is the growth rate of per capita income. In this, our results square well with the ‘legal-based’ view espoused more recently by LaPorta et al. (1997, 1998) and for which Levine (2000) finds strong cross-country evidence. When legal and financial institutions are especially underdeveloped, agency problems are severe; a financial sector, if it exists, is primitive and allows for limited amount of borrowing and lending. Much of economic production will be in activities like peasant agriculture and cottage-industry production, activities that do not rely upon financial markets but also yield low-returns. Although neither a bank-based nor a market-based system is specifically better for growth, our model suggests some advantages to having a bank-based system. In particular, since bank monitoring substitutes for entrepreneurial wealth, it enables all modern-sector firms to make larger investments than is possible under purely unintermediated finance. It also lowers the minimum entrepreneurial wealth required to obtain external finance so that the traditional sector is smaller under a bank-based system. Hence, even when a bank-based and a market-based economy grow at similar rates and have similar wealth distributions, per capita GDP in the former is permanently higher. Our model predicts rising inequality along the economy’s long-run growth path: income levels in the traditional sector stagnate, while the modern-sector enjoys sustained improve- ments in living standards. Therefore, to address equity concerns, we consider the role of policy interventions. Financial and legal reforms which reduce agency problems make it eas- 7The role of initial wealth for financing choices is particularly relevant in light of evidence that even in developed countries approximately 70% of new investment in physical capital is financed out of retained earnings (Mayer, 1988). See Allen and Gale (2001) for more recent evidence. 4 ier for modern-sector entrepreneurs to borrow. This raises the investment rate, and hence GDP growth, under both types of financial system. However, in a bank-based system, these reforms also have a level effect on per capita income. By lowering the minimum wealth needed to raise external finance, they assist some traditional sector entrepreneurs to enter the mod- ern sector — per capita GDP rises as more entrepreneurs switch from home production to the manufacture of high-productivity capital goods. In contrast, reforms in a market-based system may leave the traditional sector relatively worse-off unless they specifically reduce the costs of bank intermediation. In both regimes, a temporary income redistribution has permanent effects on the level of per capita income and on income distribution. The paper is organized as follows. We lay out the structure of the economy in Section 2. Sections 3 and 4 discuss financing options that an entrepreneur faces and her optimal investment decision. In Section 5 we characterize the balanced growth path for the economy. We discuss implications of the model and effects of policies in Section 6. Section 7 concludes. 2. The Environment Time is discrete, continues forever and is indexed by t =0, 1, 2,...,infinity. A continuum of two-period lived agents are born every period. These agents are of two types: a fraction µ of them are working households, the remaining ones entrepreneurs. Without loss of generality, let us normalize µ to 1/2 and the measure of households and entrepreneurs each to unity. There is no population growth; each agent gives birth to one offspring at the end of her youth. Economic activity encompasses a final goods sector that produces the unique consumption good, a capital goods sector that supplies inputs to final goods producing firms, and a financial sector that channels funds from lenders to borrowers. 2.1. Economic Agents A household is born with one unit of labor time in youth which it supplies inelastically to the labor market. A generation-t household’s lifetime utility depends only upon ‘old-age’ consumption. This means the entire wage income, wt, is saved. Households are the natural lenders in this economy. They invest their savings on the financial market earning a (gross) return Rasteriskmath. An entrepreneur is also born with one unit of labor time in youth that she uses to operate either of two types of technologies. Using a modern technology she can convert units of the 5 final good into a marketable capital good. Or else, she can engage in non-marketed cottage- industry or household production of the final good. Entrepreneurs are altruistic, deriving utility from their old-age consumption and the amount of bequests they leave to their offspring. A typical generation-t entrepreneur’s pref- erences are given by the ‘warm-glow’ (Galor and Zeira, 1993) utility function: UEt = beta ln cEt+1 +(1-beta)lnbt+1,betaelement (0, 1), (2.1) where bt+1 denotes bequests made. Thus, an entrepreneur also has a wealth-endowment in youth that she receives as bequest. Preferences for households and entrepreneurs are posited to be different for a simple reason. We shall shortly identify each entrepreneur with a capital good producing ‘firm’. The bequest motive in (2.1) essentially captures the continuity of each such firm in a dy- namic production economy. Altruism among households can be readily incorporated without qualitatively altering any of our basic results. Let us index an entrepreneur by j element [0, 1] and denote her initial wealth at date-t by bjt . Wealth distribution among generation-t entrepreneurs is described by the cumulative distribution function Gt(b), indicating the proportion of them with wealth less than b. Given logarithmic preferences, entrepreneur-j bequeaths a constant proportion of her realized old- age income zjt+1: bjt+1 =(1-beta)zjt+1. (2.2) Given an initial wealth distribution G0 and entrepreneurial income {zjt+1}infinityt=0, equation (2.2) tracks the wealth distribution through time.8 2.2. Production Technologies Final Goods Sector Competitive firms produce the final consumption good combining raw labor with capital goods. The underlying private technology is constant returns in capital and labor inputs: Yt = AtN1-alphat bracketleftbiggintegraldisplay jelementEt Kjt dGt bracketrightbiggalpha , 0 bj, she has to raise the deficit from the financial sector. All entrepreneurs produce the same type of capital good and are price takers. The common return they earn from renting out their capital is rho, the marginal product of capital in a competitive equilibrium and given by (2.5). For simplicity, we assume that capital goods fully depreciate upon use. 7 Cottage Industry Production Entrepreneurs also possess a technology whose output is not marketed and is entirely self- consumed (household production). As in Banerjee and Newman (1993), we identify these entrepreneurs with self-sufficient peasants and cottage industries. Such home production is assumed not to be a part of the national income accounts. Low-productivity cottage industry technology enables an entrepreneur to produce, with a one period lag, the same consumption good that the final goods sector manufactures: xt+1 = abdeltat , (2.6) where a>0 and delta element (0, 1). The entrepreneur’s choice of technology depends upon which one gives her a higher income and whether or not she is able to obtain external finance to operate the modern technology. We discuss this in details in the next section. 2.3. The Moral Hazard Problem We motivate the existence of financial markets and intermediaries by introducing agency problems in firm borrowing. Specifically, following Holmstrom (1996) and Holmstrom and Tirole (1997), we allow an entrepreneur to choose between three types of investment projects. These projects differ in their probability of success and the amount of private benefit they bring to the entrepreneur. Suppose the entrepreneur raises funds amounting to qjt >bjt for her investment. When the project succeeds, it realizes a verifiable amount of capital, Kjt+1 = qjt . (2.7) But should it fail, it produces nothing. The moral hazard problem arises from the fact that the probability of success depends on an unobserved action taken by the entrepreneur. The unobserved action can be interpreted as her choice on how to spend qjt . She can spend it on an efficient technology that results in success for sure, but uses up all of qjt . Or, she can spend it on one of two inefficient technologies that may not succeed. One of these technologies, a low moral hazard project, costs qjt -vqjt ,leavingvqjt for the entrepreneur to appropriate. The other inefficient choice, a high moral-hazard project, costs qjt - Vqjt which leaves Vqjt in private benefits to the entrepreneur. Both inefficient technologies carry the same probability of success, pi,butwe assume that 0 0 while uninformed investors are paid thetaUt+1 > 0, where thetaEt+1 + thetaUt+1 = rhot+1qt. Entrepreneur-j prefers to be diligent, that is, invests qt in the efficient technology as long as she earns an incentive compatible return: thetaEt+1 greaterequal pithetaEt+1 + Vqt arrowdblrightthetaEt+1 greaterequal V1 -piqt. (3.1) This, of course, implies that outside investors get paid at most thetaUt+1 = bracketleftbigrhot+1 -V/(1 -pi)bracketrightbig qt. This is the pledgeable expected income that firms can credibly commit to their investors. The household participation constraint requires this income not to be lower than the amount that can be earned on the international capital market, Rasteriskmath bracketleftbigqt -bjt bracketrightbig. 10 Hence, a necessary and sufficient condition for the entrepreneur to have access to direct finance is that bracketleftbigg rhot+1 - V1 -pi bracketrightbigg qt greaterequal Rasteriskmath bracketleftbigqt -bjt bracketrightbig . Defining bHt (qt) equivalence bracketleftbigg 1 - 1Rasteriskmath parenleftbigg rhot+1 - V1 -pi parenrightbiggbracketrightbigg qt, (3.2) we conclude that only entrepreneurs with bjt greaterequal bHt (qt) are able to obtain direct finance if they want to invest qt. 3.2. Indirect Finance For indirect or intermediated finance, there are three parties to the financial contract: the entrepreneur, the bank and uninformed investors. As before, an optimal contract requires that no party earns anything when the project fails. When it succeeds, the total return, rhot+1qt, is distributed such that thetaEt+1 + thetaUt+1 + thetaBt+1 = rhot+1qt, with thetaB denoting the bank’s returns. If banks choose to monitor firms, a generation-t entrepreneur’s incentive constraint be- comes thetaEt+1 greaterequal pithetaEt+1 + vqt arrowdblrightthetaEt+1 greaterequal v1 -piqt. (3.3) Monitoring occurs the same period that banks lend to firms. Banks have an incentive to monitor firms since it reduces the incentive compatible payoff that has to be committed to an entrepreneur, and thereby raises banking revenues. But monitoring is also costly and it reduces the total amount of loans a bank can make. We assume that banks discount monitoring costs at their opportunity cost, Rasteriskmath. Banks are then willing to monitor only if they earn at least as much under monitoring as they would (in an expected sense) under no-monitoring thetaBt+1 -Rasteriskmathgammaqt greaterequal pithetaBt+1 arrowdblrightthetaBt+1 greaterequal R asteriskmathgamma 1 -piqt. (3.4) The pledgeable expected income in this case is bracketleftbigg rhot+1 - v + gammaR asteriskmath 1 -pi bracketrightbigg qt. Let Ljt be the amount that a bank lends to entrepreneur-j. The bank’s return from this entrepreneur is then thetaB,jt+1 = RLt+1Ljt , 11 where RLt+1 is the (gross) loan rate charged to borrowers. Note that the lending rate is the same across borrowers as both lenders and borrowers operate in competitive markets. Since bank finance is relatively more expensive, entrepreneurs accept only the minimum amount necessary. From (3.4) this implies Ljt (RLt+1,qt)= gammaR asteriskmath (1 -pi)RLt+1qt. (3.5) Uninformed investors must supply the balance Mjt = qt -bjt -Ljt , whenever this amount is positive. Therefore, a necessary and sufficient condition for entrepreneur-j to be financed is bracketleftbigg rhot+1 - v + gammaR asteriskmath 1 -pi bracketrightbigg qt greaterequal Rasteriskmath bracketleftbigqt -bjt -Ljt bracketrightbig . We can rewrite this condition as bjt greaterequal bLt (RLt+1,qt) equivalence qt -Ljt (RLt+1,qt) - qtRasteriskmath bracketleftbigg rhot+1 - v + gammaR asteriskmath 1 -pi bracketrightbigg . Only entrepreneurs with bjt greaterequal bLt (RLt+1,qt) are able to convince uninformed investors to supply enough funds for a project of size qt. 3.3. The Bank’s Problem Given the loan demand Ljt from each borrowing firm j from (3.5) above, the aggregate demand for bank loans is Lt = integraldisplay jelementEBt Ljt dGt = bracketleftbigg gammaRasteriskmath (1 -pi)RLt+1 bracketrightbigg integraldisplay jelementEBt qjt dGt, where EBt propersubset Et denotes the subset of firms borrowing from banks in period t. Hence, the total monitoring cost borne by the banking sector is integraldisplay jelementEBt gammaqjt dGt = (1 -pi)R L t+1Lt Rasteriskmath . Let Dt denotes the flow of deposits into the banking sector. Then banking profits in period-(t +1)are given by PiBt+1 = RLt+1Lt -RasteriskmathDt. (3.6) Banks face the resource constraint that total loans cannot exceed total deposits net of mon- itoring costs: Lt <=< Dt - integraldisplay jelementEBt gammaqjt dGt. (3.7) 12 The banking sector’s optimization problem in period t is to choose Lt so as to maximize PiBt+1 subject to the constraints (3.3), (3.4) and (3.7). Since bank profits are increasing in total loans, (3.7) holds with equality: Lt = Dt - integraldisplay jelementEB,t gammaqjt dGt = Dt - bracketleftbigg(1 -pi)RL t+1 Rasteriskmath bracketrightbigg Lt. (3.8) Moreover, in a competitive equilibrium, the banking sector earns zero profits. From (3.6) we then have RLt+1Lt = RasteriskmathDt, (3.9) It follows from equations (3.8) and (3.9) that Lt = piDt, and RLt+1 = R asteriskmath pi . (3.10) Hence, using (3.5), we observe that phijt equivalence L j t qt = gamma parenleftbigg pi 1 -pi parenrightbigg . (3.11) In other words, for all j element EBt , banks finance a fixed proportion of the firm’s investment, irrespective of entrepreneur-j specific characteristics, that is, bjt . Note that in order that the loan size does not exceed investment size, that is phijt <=< 1, monitoring costs should not be so high as to make it impossible for bank intermediation to resolve moral hazard problems. Hence, we restrict monitoring cost such that: gamma <=< (1 -pi)/pi. (Assumption 1) Taking into account the optimal loan size (3.11), for any qt, the lower wealth cut-off becomes bLt (qt)= bracketleftbigg 1+gamma - 1Rasteriskmath parenleftbigg rhot+1 + v1 -pi parenrightbiggbracketrightbigg qt. (3.12) It is natural to assume that bHt (qt) greaterequal bLt (qt), or else there will be no demand for intermediation — monitoring would be too costly to be socially useful. From (3.2) and (3.12), we have bHt (qt) greaterequal bLt (qt) as long as the expected gains from monitoring exceed its cost: V -v 1 -pi greaterequal gammaR asteriskmath. (Assumption 2) 13 3.4. Entrepreneur’s Income under Optimal Contracts Denote entrepreneur-j’s income in the second period of life by zjt+1. Each entrepreneur has the option of investing her funds on the domestic financial sector or lending abroad, both of which would earn her a gross return Rasteriskmath. Consider the case of an entrepreneur-j who does not qualify for any external finance, that is, her internal funds are too small, bjt <=< bLt (qt). Clearly, she would prefer to engage in household production instead of lending out her funds iff abdeltat greaterequal Rasteriskmathbt arrowdblrightbt <=bjt and the wealth distribution Gt(b), (i) Entrepreneurs with bjt bL(qt) for any qt (by Assumption 2), the bH(qt) line is steeper than bL(qt). The intersection points of these rays with qasteriskmath are labeled basteriskmathH and basteriskmathL respectively. Suppressing the j-superscript, consider a generation-t entrepreneur with inherited be- quest level bt >basteriskmathL. Note that this basteriskmathL is the same as (4.1), obtained by substituting qt = qasteriskmath into (4.3). In Figures 1(a) and 1(b), qI,t and qU,t are given by the points of intersection of bL(qt) and bH(qt) with the entrepreneur’s wealth, bt. Observe that bt qI,t. Such an entrepreneur desiring to invest more than qI,t cannot convince uninformed investors to supply enough funds for her project, and is completely rationed from the credit market. She can only resort to household production in that case and earn an income zt+1 = abdeltat (from (3.14)). This earning is given by the horizontal line PQ in Figures 1(a) and (b). Similarly, bL(qt) <=< bt 0 and bH(qt) > 0 for any qt. 17 [alphaA -(1 + gamma)Rasteriskmath] qt, shown by the flatter line HN with intercept Rasteriskmathbt in Figures 1(a)and (b). Finally, for any qasteriskmath <=< qt <=< qU,t,bt greaterequal bH(qt). An entrepreneur can fund an investment in this range by raising funds directly from the market without requiring any bank finance. His earning zt+1 = Rasteriskmathbt +[alphaA -Rasteriskmath] qt is the steeper line EF with intercept Rasteriskmathbt in Figures 1(a) and (b). To summarize, when an entrepreneur chooses her level of investment, her earnings are given as: zt+1(qt| bt)= bracelefttpbraceex braceexbraceleftmid braceexbraceexbraceleftbt Rasteriskmathbt + qt(alphaA -Rasteriskmath), if qt element [qasteriskmath,qU,t] Rasteriskmathbt + qt [alphaA -(1 + gamma)Rasteriskmath] , if qt element (qU,t,qI,t] abdeltat , if qt element (qI,t, infinity). An entrepreneur chooses qt so as to maximize zt+1(qt). Solving this optimization problem is straightforward from Figure 1. Income zt+1(qt) is given by the piecewise linear schedule EF, HN and PQ. Two possibil- ities arise: qI,t is the investment choice when the height of the point N is greater than that of the point F [Figure 1(a)], whereas qU,t is chosen when the opposite holds [Figure 1(b)].15 Closed-form solutions for these investment levels can be obtained using (4.3) and (4.4). In particular, qI,t is the solution to the equation bt = bL(qt) so that qI,t = bracketleftbigg Rasteriskmath v/(1 -pi) -[alphaA -(1 + gamma)Rasteriskmath] bracketrightbigg bt. (4.5) Similarly, qU,t solves bt = bH(qt) which gives us qU,t = bracketleftbigg Rasteriskmath V/(1 -pi) -(alphaA -Rasteriskmath) bracketrightbigg bt. (4.6) Since zt+1(qt) is strictly increasing in the range qt element [qasteriskmath,qU,t], maximal earning occurs at qt = qU,t and is given by zt+1(qU,t)= bracketleftbigg [V/(1 -pi)] Rasteriskmath V/(1 -pi) -(alphaA -Rasteriskmath) bracketrightbigg bt. (4.7) Likewise, since zt+1(qt) is strictly increasing in the range qt element (qU,t,qI,t], the maximum earning in this range occurs at qt = qI,t, and the maximum earning is zt+1(qI,t)= bracketleftbigg [v/(1 -pi)] Rasteriskmath v/(1 -pi) -[alphaA -(1 + gamma)Rasteriskmath] bracketrightbigg bt. (4.8) 15In Figure 1(a), qI,B denotes bank borrowing, qI,M denotes market borrowing and qI,S denotes self- financing in a bank-based system. Similarly for a market-based system in Figure 1(b). 18 It follows that an entrepreneur chooses qI,t over qU,t iff, zt+1(qI,t) greaterequal zt+1(qU,t) arrowdblboth alphaA -(1 + gamma)R asteriskmath v greaterequal alphaA -Rasteriskmath V , (4.9) a condition that does not depend upon borrower characteristics, that is, on bt. Note that incentive constraints of all modern-sector entrepreneurs are binding, since qI,t (qU,t)isgiven by the intersection of bt with bL(qt) [bH(qt)]. 4.3. Implications for Financial System Consider now the financial system resulting from firm-financing decisions. As long as (4.9) holds [Figure 1(a)], except for the fraction G0(basteriskmathL) of entrepreneurs who are credit- constrained, all capital goods producers finance their investment through a mix of interme- diated and unintermediated finance. We label this a bank-based financial system. On the other hand, if (4.9) does not hold [Figure 1(b)], unconstrained entrepreneurs earn a higher income with purely unintermediated finance. Despite this, one group of en- trepreneurial families have to rely upon bank-finance, at least in the short-run. To see this, consider entrepreneurs with basteriskmathL <=< bt 0 and partialdiffbasteriskmathL/partialdiffv > 0. 24Recall that along the balanced-growth path, incomes between modern-sector and traditional-sector en- trepreneurs are diverging. 28 tems. This is especially true in environments where setup costs are large relative to average wealth levels and when a more equitable income distribution is of paramount concern. Redistributive Policies Given that the traditional sector stagnates even as the modern-sector enjoys sustained income gains, a natural question to ask is whether policies could ameliorate this widening inequality. In a bank-based system, one possibility is to tax wealthier (unconstrained) en- trepreneurs and use the proceeds to subsidize the banking sector. For instance, the subsidy may be provided every time a bank spends resources monitoring a borrowing firm. This is equivalent to a reduction in gamma, the cost that is directly borne by banks. The outcome is a lower basteriskmathL and a larger proportion of entrepreneurs in the modern-sector. How long it takes for such a policy to pull out the entire traditional sector depends upon the initial size of this sector, and how high agency costs are. Moreover, subsidies to the banking sector will work well only if there are no agency problems within the banks, an assumption we have implicitly maintained in our analysis. If that were not the case, such subsidies could result in worse investment outcomes and efficiency loss. A more desirable policy intervention may be to directly redistribute wealth from wealth- ier entrepreneurs to poorer ones. The effect on wealthier entrepreneurs would be similar to that under the previous policy, but providing additional wealth to credit-constrained en- trepreneurs may enable all of them to eventually overcome basteriskmathL. Such a policy has the added benefit of working in a bank-based economy as well as a market-based one. As in Banerjee and Newman (1993), neither type of policy intervention has to be perma- nent. Temporary subsidies enable constrained entrepreneurs to overcome basteriskmathL. A subsequent withdrawal of the subsidy does not push these entrepreneurial families back into the tradi- tional sector as long as they have accumulated wealth beyond basteriskmathL while the policy was in place. On balance, therefore, we conclude that although there may not be distinct growth ad- vantages to having a particular financial regime, bank-based systems have an edge along other dimensions. Intermediated finance confers certain benefits for economic development. Two different financial structures may lead to similar growth rates, but a bank-based system has a level effect on per capita income and leads to a faster structural transformation. More- over, developing countries contemplating financial sector reforms to reduce agency problems in the loanable funds market will obtain higher economic payoffs under bank-based systems due to the structural transformation that results. Our analysis is, thus, complementary to 29 some recent contributions, notably by Rajan and Zingales (1998b, 1999) and Tadesse (2001), which make a strong case for bank-based systems in developing countries. 7. Conclusion The chief contribution of this paper has been to shed light on the ongoing debate about the relative merits of bank-based and market-based financial systems for growth and de- velopment. Many developing countries have been moving towards market-based systems in recent years without a clear consensus that such systems are necessarily better. Hence, it is important that growth theory addresses this debate to better inform policy-making. From a growth perspective, we do not find that one type of system is invariably better than the other. Indeed, it is quite possible for two types of systems in two different countries to deliver similar growth rates of per capita GDP. Moreover, and consistent with recent cross- country evidence, we argue that the quality of a country’s financial and legal institutions are more important for its growth. But bank-based systems have some advantages over those that are market-based. For one, levels of investment and per capita GDP are higher under a bank-based system. Bank monitoring resolves some of the agency problems and enable firms to borrow more. Arms- length market finance plays no such role and results in a lower amount of external finance available to all firms. Secondly, bank-based systems allow greater participation in manufacturing activities, by providing external finance to a larger number of entrepreneurs. The implication is that the traditional sector is smaller and wealth distribution better under such a system. Policy reforms in a bank-based system raise the growth rate and reduce the size of the traditional sector; in a market-based system they improve growth but leave the traditional sector unaffected unless such policies also reduce the costs of intermediated finance. A temporary redistribution policy, under both financial regimes, has permanent effects on growth, distribution and relative size of the traditional sector. 30 8. Appendix A.1. Positive Growth Rates In the intermediated finance regime the growth rate gI is defined by 1+gI equivalence (1 -beta) bracketleftbigg [v/(1 -pi)]Rasteriskmath v/(1 -pi) -[alphaA -(1 + gamma)Rasteriskmath] bracketrightbigg . From Assumption 4 we have v/(1 -pi) <=< alphaA -R asteriskmath(1 + gamma) 1 -(1 -beta)Rasteriskmath arrowdblright1 - alphaA -R asteriskmath(1 + gamma) v/(1 -pi) <=< (1 -beta)R asteriskmath arrowdblright(1 -beta) bracketleftbigg [v/(1 -pi)]Rasteriskmath v/(1 -pi) -[alphaA -(1 + gamma)Rasteriskmath] bracketrightbigg greaterequal 1. It follows that gI greaterequal 0. Similarly, using Assumption 5, it can be shown that the growth rate under the uninter- mediated finance regime, gU greaterequal 0. A.2. Investment Size in Bank-based and Market-based Systems Here we consider if two countries that share the same growth rate but under different financial systems also invest equal amounts. Take two countries, 1 and 2 where 1 uses mixed finance, but 2 uses unintermediated finance. Suppose both have the same growth rate so that alphaA -(1 + gamma1)Rasteriskmath v1/(1 -pi) = alphaA -Rasteriskmath V2/(1 -pi). (8.1) We shall compare the size of investment any particular entrepreneur with internal funds bj makes in each of these countries. We have qj1 greaterequal qj2 arrowdblboth (1 + gamma1) - 1Rasteriskmath (alphaA - v11 -pi) <=< 1 - 1Rasteriskmath (alphaA - V21 -pi) arrowdblboth V2 -v11 -pi greaterequal gamma1Rasteriskmath. (8.2) 31 From (8.1), v1 V2 = alphaA -(1 + gamma1)Rasteriskmath alphaA -Rasteriskmath =1- gamma1Rasteriskmath alphaA -Rasteriskmath arrowdblright 1 - v1V 2 = gamma1R asteriskmath alphaA -Rasteriskmath arrowdblright V2 -v11 -pi = V21 -pi gamma1R asteriskmath alphaA -Rasteriskmath . Substituting this into the LHS of (8.2), V2 1 -pi gamma1Rasteriskmath alphaA -Rasteriskmath greaterequal gamma1R asteriskmath arrowdblboth V21 -pi greaterequal alphaA -Rasteriskmath. Now from the participation constraint of the entrepreneur in country 2, zj =(alphaA -Rasteriskmath)qj2 + Rasteriskmathbj = V21 -piqj2, we have, V 2 1 -pi >alphaA-R asteriskmath as long as bj > 0. Hence, investment size in country 2 is smaller for every entrepreneur in the modern-sector, i.e., qj1 >qj2. Consequently, per capita GDP in country 1 is larger, i.e., Y1 >Y2. A.3. basteriskmathL in Bank-based and Market-based Systems We show here that basteriskmathL is lower for a bank-based economy than a market-based one, if both have the same growth rate. Once again consider country 1 with a bank-based system and country 2 with a market-based one and, g1 = g2. We need to show that basteriskmathL,1 alphaA -Rasteriskmath V1/(1 -pi), and alphaA -Rasteriskmath(1 + gamma2) v2/(1 -pi) < alphaA -Rasteriskmath V2/(1 -pi). Also, g1 = g2 implies that alphaA -Rasteriskmath(1 + gamma1) v1/(1 -pi) = alphaA -Rasteriskmath V2/(1 -pi). 32 Combining these three relations we have the following inequality: alphaA -Rasteriskmath(1 + gamma1) v1/(1 -pi) > alphaA -Rasteriskmath(1 + gamma2) v2/(1 -pi) arrowdblright v2 -v11 -pi > v11 -pi R asteriskmath(gamma1 -gamma2) alphaA -Rasteriskmath(1 + gamma1). (8.3) Now, from entrepreneur-j’s incentive constraint in country 1,wehave zj1 =[alphaA -(1 + gamma1)Rasteriskmath]qj1 + Rasteriskmathbj = v11 -piqj1, so that, for any bj > 0, v 1 1 -pi >alphaA-(1 + gamma1)R asteriskmath. Combining this with (8.3) above, we get v2 -v1 1 -pi >R asteriskmath(gamma 1 -gamma2). (8.4) To have basteriskmathL,1 Rasteriskmath(gamma1 -gamma2), which holds by (8.4). 33 References [1] Allen, Franklin and Douglas Gale (2000), Comparing Financial Systems, Cambridge, MA, MIT Press. [2] Allen, Franklin and Douglas Gale (2001), “Comparative Financial Systems: A Survey”, mimeo, New York University. [3] Arrow, Kenneth J. (1962), “The Economic Implications of Learning by Doing,” Review of Economic Studies, vol. 29, pp. 155-73. [4] Banerjee, Abhijit and Andrew Newman (1993), “Occupational Choice and the Process of Development,” Journal of Political Economy, vol. 101, pp. 274-98. [5] Bernanke, Ben and Mark Gertler (1989), “Agency Costs, Net Worth and Business Fluc- tuations”, American Economic Review, vol. 79, pp. 14-31. [6] Bernanke, Ben, Gertler, Mark and Simon Gilchrist (2000), “The Financial Accelerator in a Quantitative Business Cycle Framework”, in The Handbook of Macroeconomics, North-Holland. [7] Bencivenga, Valerie and Bruce Smith (1991), “Financial Intermediation and Endogenous Growth”, Review of Economic Studies, vol. 58, pp. 195-209. [8] Boot, Arnoud W. A. and Anjan V. Thakor (1997), “Financial System Architecture,” Review of Financial Studies, vol. 10, pp. 693-733. [9] Boyd, John H. and Edward Prescott (1986), “Financial Intermediary-Coalitions,” Jour- nal of Economic Theory, vol. 38, pp. 211-32. [10] Boyd, John H. and Bruce D. Smith (1998), “The Evolution of Debt and Equity Markets in Economic Development,” Economic Theory, vol. 12, pp. 519-60. [11] de la Fuente, Angel and Jose Maria Marin (1996), “Innovation, Bank Monitoring, and Endogenous Financial Development”, Journal of Monetary Economics, vol. 38, no. 2, pp. 269-301. [12] Diamond, Douglas (1984), “Financial Intermediation and Delegated Monitoring,” Re- view of Economic Studies, vol. 51, pp. 393-414. 34 [13] Diamond, Douglas (1991), “Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt”, Journal of Political Economy, vol. 99, pp. 689-721. [14] Galor, Oded and Joseph Zeira (1993), “Income Distribution and Macroeconomics,” Review of Economic Studies, vol. 60, pp. 35-52. [15] Gerschenkron, Alexander (1962), Economic Backwardness in Historical Perspective, Cambridge, MA: Harvard University Press. [16] Goldsmith, Raymond (1969), Financial Structure and Development, Yale University Press, New Haven, CT. [17] Greenwood, Jeremy and Boyan Jovanovic (1990), “Financial Development, Growth and the Distribution of Income”, Journal of Political Economy, vol. 98, no. 5, pp. 1076-107. [18] Greenwood, Jeremy and Bruce Smith (1997), “Financial Markets in Development, and the Development of Financial Markets,” Journal of Economic Dynamics and Control, vol. 21, pp. 145-81. [19] Gurley, John and E. Shaw (1955), “Financial Aspects of Economic Development,” American Economic Review, vol. 45, pp. 515-38. [20] Hellwig, Martin (1991), “Banking, Financial Intermediation and Corporate Finance”, in European Financial Integration, (eds.) Alberto Giovannini and Colin Mayer, Cambridge University Press, U.K. [21] Hicks, John (1969), A Theory of Economic History, Oxford, UK, Clarendon Press. [22] Holmstrom, Bengt (1996), “Financing of Investment in Eastern Europe,” Industrial and Corporate Change, vol. 5, pp. 205-37. [23] Holmstrom, Bengt and Jean Tirole (1997), “Financial Intermediation, Loanable Funds, and the Real Sector,” Quarterly Journal of Economics, vol. 112, pp. 663-91. [24] La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert Vishny (1997), “Legal Determinants of External Finance,” Journal of Finance, vol. 52, pp. 1131-50. [25] La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert Vishny (1998), “Law and Finance,” Journal of Political Economy, vol. 106, pp. 1113-55. 35 [26] Levine, Ross (1997), “Financial Development and Economic Growth: Views and Agenda”, Journal of Economic Literature, vol. 35, pp. 688-726. [27] Levine, Ross (2000), “Bank-based or Market-Based Financial Systems: Which is Bet- ter?”, mimeo, University of Minnesota. [28] Mayer, Colin (1988), “New Issues in Corporate Finance”, European Economic Review, vol. 32, pp. 1167-89. [29] North, Douglass C. (1981), Structure and Change in Economic History,NewYork,W. W. Norton. [30] Rajan, Raghuram and Luigi Zingales (1998a), “Financial Dependence and Growth”, American Economic Review, vol. 88, pp. 559-86. [31] Rajan, Raghuram and Luigi Zingales (1998b), “Which Capitalism? Lessons from the East Asian Crisis,” Journal of Applied Corporate Finance, vol. 11, no. 3, pp. 40 - 48. [32] Rajan, Raghuram and Luigi Zingales (1999), “Financial Systems, Industrial Structure and Growth”, mimeo, University of Chicago. [33] Romer, Paul (1986), “Increasing Returns and Long-Run Growth,” Journal of Political Economy, vol. 92, pp. 1002-37. [34] Scharfstein, David (1988), “The Disciplinary Role of Takeovers”, Review of Economic Studies, vol. 55, pp. 185-99. [35] Tadesse, Solomon (2001), “Financial Architecture and Economic Performance: Interna- tional Evidence”, mimeo, University of South Carolina. [36] Williamson, Stephen (1986), “Costly Monitoring, Financial Intermediation and Equi- librium Credit Rationing”, Journal of Monetary Economics, vol. 18, pp. 159-179. [37] Williamson, Stephen (1987), “Financial Intermediation, Business Failures, and Real Business Cycles,” Journal of Political Economy, vol. 95, no. 6, pp. 1196-1216. 36 450 qU,t zt+1 b q L ( )t b q( )t 0 qtq I,t bt abt δ [1 ] - γ ( π / 1 − π ) q tqI,B,t+1 qI,M,t+1 qI,S,t+1 E F H N PQ *q * Hb * Lb * tRb Figure 1(a): Investment Choice in a Bank-based System i qU,t zt+1 b q L ( )t b q( ) 0 qtq I,t bt abt δ qU,M,t+1 qU,S,t+1 E F H N PQ 450 * tRb * Hb * Lb *q Figure 1(b): Investment Choice in a Market-based System ii