Does the removal of intra-state entry barriers increase welfare? Will all banks survive? Will it lead to a consolidation of the banking industry? The experience of credit market deregulation has not been always successful. Credit market liberalisation, via the removal of entry barriers, of limitations of activity and of markets for funding is generally recognised at the origin of banking crises like the American Savings & Loans, the Scandinavian countries’ banking crises at the end of the 80’s, the Japanese crises not yet resolved and more recently and on a large scale the East-Asian countries. However there are doubts about whether these problems arose because of liberalisation per se or rather because of failure of prudential regulation in disciplining banks whose lenders were protected by deposit insurance and State guarantees. Financial markets integration in Europe is still at too an early stage to evaluate, but certainly the very clear process that has emerged is a consolidation of the banking system. Moreover, such a process appears to take place worldwide and somehow be the response to anticipation of future competition. Berger, Kashyap and Scalise (1995) documents a striking amount of consolidation in the American banking industry over the 1979-94 period and attribute this consolidation to the relaxation of intra-state entry barriers, moreover consolidation was favoured by regulators through the easing of the process for approving mergers. footnote Does a concentrated banking industry dominate a fragmented one? This is the question this paper seeks to address. We shall consider banks as limited-liable intermediaries that borrow form investors and lend to firms, and concentrate on bank’s (incentive-compatible) information production about borrowers, that is on bank’s (incentive-compatible) screening activity. We shall show that when perfectly diversified credit portfolios cannot be constructed, credit allocation depends on aggregate bank capital and on the number of banks that can operate in the same market. In particular: i) a concentrated banking industry, one where aggregate bank capital is held by few banks, leads to credit allocation closer to the first-best optimum; and ii) in the absence of banking industry consolidation, the removal of intra-state entry barriers reduces welfare. Moreover, not all independent banking organisations that were viable in formerly protected national markets remain so when markets are integrated. One reason that makes fully diversified credit portfolios unfeasible is macroeconomic (systematic) risk. In the economy considered in this paper, a project return depends both on the intrinsic quality of the project "firm" and on overall economic performance. A firm, regardless of its quality, is more likely to succeed in a boom than a recession, and firms that manage to succeed in a recession are the fit ones. By (privately) engaging in costly information production about firms’ quality, that is by screening loan applicants, a bank can assess the quality of borrowing firms but not its asset-portfolio return which depends also on overall economic performance. The bank pursues the interests of its shareholders (insiders), it may then have an incentive to take bad risk: it may be tempted to bet that macroeconomic factors will support firms’ performance, avoid costly search for good type firms – financing de facto negative net-present value projects – and shift resulting losses on (outsiders) investors. At equilibrium projects that are undertaken have positive net-present value (that is, agents’ participation constraints are satisfied): banks that are active, find it profitable to engage in information production about borrowers. The questions are: (i) what is a bank’s incentive constraint , (ii) how is this affected by banking industry and market structures, and what the implications on credit allocation. The analysis of a bank’s profit maximising choice provides the answer to question (i). The bank finds it incentive-compatible to engage in information gathering (that is search for good-type firms) only if the amount of capital that it invests in its asset portfolio does not fall below a threshold level which is decreasing in its profit margin (the wedge between bank’s lending and borrowing rates) and increasing in the cost of searching for good type firms. The lower the average quality of loan applicants, the higher this cost. The bank’s incentive-compatibility constraint thus sets a ceiling on the amount of outside financing that the bank can raise and hence on the amount of lending it can make. The higher bank capital and the profit margin and the better the quality of its loan applicants, the higher its lending volume. With regard to (ii), the key observation is that when several banks are active in the same market, a rejected borrower does not exit he applies to a second bank and if rejected he tries again with a third bank.... The screening and re-application process acts to worsen banks’ pool of applicants, it then acts to increase banks’ costs of searching for good-type borrowers and thereby to lower the amount of (incentive-compatible) lending, for any given amount of bank capital. The higher the number of banks that can operate in the same market, either because the banking industry is more fragmented and/or intra-state entry barriers are removed, the higher the short-fall of aggregate lending from the first-best optimum and the lower the average quality of borrowing firms. Moreover, at equilibrium, an active bank (that is a bank that engages in lending) is necessarily viable, i.e. it finances firms whose projects have positive net-present value. This sets an upper bound on the number of banks that are active at equilibrium: banks that would be viable in protected national (or local) markets are non-viable when intra-state entry barriers are removed. These results follow from two crucial observations: (i) the information externality of screening and its adverse pool effects; (ii) the size of a bank (the amount of lending it undertakes) is endogenously constrained: it cannot exceed a ceiling positively linked to its capital, to its (endogenous) profit margin and to the quality of its loan applicants, where the latter depends on banking industry and market structures. Observation (i) has been made and explored before by Broecker (1990), Riordan (1993) and Nakamura (1996) for given statistical properties’ of banks’ screening technologies, and by Gehrig (1997) for endogenous screening intensity from a continuum of potential levels. These papers abstract from banks’ agency problems by assuming that banks finance firms with their own (unlimited) capital. The size of a bank is unconstrained and the welfare effects of competition depend on the specificity of the model with regard to whether lenders can react to each other (Broecker (1990)) and whether they observe a continuous signal and base their lending decisions on a chosen signal’s cut-off (Riordan (1993)); Gehrig (1997) provides conditions under which screening efforts are reduced by competition. Observation (ii) is specific to this paper, it follows from banks being limited liable and subject to delegation – agency – problems. It provides a role for banking industry structure, independent of competition issues. We do indeed find that banks compete more aggressively — banks’ profit margins shrink — when the industry structure is more fragmented. This reinforces the conclusion that aggregate lending and its average quality are lower with a fragmented industry structure than with a concentrated one. Banks’ agency problems have been examined earlier. The financial intermediation literature has stressed the role of risk diversification in solving the conflict of interests between the bank and investors and noted that if all risk could be diversified, agency problems would disappear (Diamond 1984; Ramakrishnan and Thakor 1984; Bhattacharya and Thakor 1993). Boot and Greenbaum (1993) suggests that bank’s reputational concerns could ameliorate moral hazard problems and substitute rents in incentivating monitoring of ongoing projects. Bernanke and Gertler (1987) are the first to point out that bank capital affects the scale of banking. In their model, investors’ payoffs cannot be conditioned on banks’ risky asset returns since these are not public information, bank capital (perfectly) collateralises bank liabilities. Chiesa (1997) analyses banks’ agency problems and regulatory issues when risk diversification is limited and banks monitor firms’ choices of projects. In contrast to the previous literature, this paper concentrates on banks’ ex ante information production. The paper is structure as follows. Section 2 presents the model. Section 3 derives the incentive-compatibility constraint for bank information production in the (simplest) case of a single bank. Section 4 analyses credit allocation in a one-bank economy. This can be one where there is a single (monopoly) bank, or, equivalently, one with several (monopoly) banks operating in protected national – local – markets. Section 5 analyses credit allocation in a two-banks economy. This economy differs from the previous one either because the banking industry is more fragmented, or equivalently because intra-state entry barriers are removed. Section 6 concludes.