European integration, the emergence of Eastern Europe and increased global competition have brought about rapid changes in financial and real markets which call for new analytical tools and ideas, to which both industrial organization and finance have made important contributions. Tools and ideas from industrial organization are permeating finance analysis while developments in finance are increasingly being applied to industrial economics. A joint conference with the Fundación Banco Bilbao-Vizcaya, held in San Sebastián on 21/23 April, brought together leading economists in financial economics and industrial organization to consider such topics as the industrial organization approach to financial markets, the influence of financial structure on growth and the relationship between banks and industries. The conference was organized by Xavier Vives, Director of the Institut d'Anàlisi Econòmica (CSIC), Barcelona, and Co-Director of CEPR's Industrial Organization programme. The financial assistance and organization of this conference provided by the Fundación BBV are gratefully acknowledged.

Franklin Allen (University of Pennsylvania) presented `A Welfare Comparison of the German and U.S. Financial Systems', written with Douglas Gale, which noted that cross-country differences in financial systems and economic performance raise interesting questions concerning their causes and effects. Allen described and compared two idealized polar extremes: the `German model', in which banks and other intermediaries predominate, and the `US model', in which highly developed financial markets play the dominant role. He first considered risk sharing, noise suppression and the provision of services, noting that the US financial system lacks intergenerational risk-sharing opportunities but nevertheless provides many opportunities for cross-sectional risk sharing. He also considered the firm side of the market, focusing on the allocation of investment, the market for corporate control, monitoring and long-term relationships. He found that any welfare comparison of different financial systems must consider a wide range of factors on both the investor and firm sides of the market. The presence of imperfections ensures that it is not clear whether markets are the best institutions available.

Michael Riordan (Boston University) stressed the need for general rather than partial equilibrium analysis to compare the welfare effects of different financial systems and provide satisfactory explanations for the existence of banks. He also argued that banks and markets are complements rather than substitutes, as Allen's framework appeared to suggest.
While the adverse effects of ex post monopoly power on ex ante incentives are well documented in the industrial organization literature, their applications to financial markets have been surprisingly few. A Jorge Padilla (Centro de Estudios Monetarios y Financieros (CEMFI), Madrid, and CEPR) presented `Endogenous Communication among Lenders and Entrepreneurial Incentives', written with Marco Pagano, in which they accounted for credit registers and credit information bureaux as instruments to limit ex post monopoly power and thus improve ex ante incentives. In a two-period model with imperfectly competitive banks, heterogeneous entrepreneurs may be concerned that banks enjoying an informational monopoly about them will appropriate part of the returns to their effort now by charging higher interest rates in future. Credit registers allow banks to commit credibly to forgo such opportunistic behaviour by sharing private information on the quality of their borrowers. This raises banks' current profits by raising borrowers' effort levels, but it also reduces their future profits by promoting competition. If both good and bad entrepreneurs exist but only the former undertake profitable projects, information sharing is essential for the credit market to function at all. Padilla also outlined a simple reputation system that can prevent cheating by banks participating in the information-sharing agreement and considered the circumstances in which banks will prefer to agree to exchange only limited information about their borrowers.

Thomas Gehrig (Universität Basel and CEPR) stressed the need for further investigation of the type of information traded by credit registers to determine their nature and, more importantly, to test the rival hypothesis that sharing information on credit volumes of individuals may also enable banks to divide up their markets and engage in anticompetitive conduct.

In `The Effect of Credit Market Competition on Lending Relationships', written with Mitchell Petersen, Raghuram Rajan (University of Chicago) considered whether firms that face competitive credit markets can form strong ties to particular creditors and whether increased competitiveness diminishes the benefits to firms of so doing. He outlined a simple framework to investigate this issue empirically for small businesses in the US, which obtain external finance almost entirely from banks while the extent of banking competition varies substantially across regions. Rajan maintained that creditors in more concentrated credit markets should be more likely to finance credit-constrained firms since they will internalize the benefits of assisting them more easily, and data from the 1988-9 National Survey of Small Business Finances, covering 3,404 small firms (with fewer than 500 employees) stratified by census region, urban/rural location and employment size provided support for this hypothesis. The firm's life-cycle also played an important role: creditors charged less than the competitive interest rate when firms were young but more when they were old.
In his discussion, Rafael Repullo (CEMFI, Madrid, and CEPR) developed a two-period theoretical model of a firm seeking finance in the presence of adverse selection and moral hazard to address some related issues. He questioned Rajan's use of deposit market concentration to proxy concentration in the market for firms' loans and stressed the need for extra care in interpreting the empirical results, since they were derived from an analysis of cross-section rather than panel data.

A frequently cited benefit of borrowing from commercial banks is that it is much easier to renegotiate and restructure private debt than public debt claims in situations of financial distress. Christopher James (University of Florida) presented the paper `When Do Banks Take Equity? An Analysis of Bank Loan Restructurings and the Role of Public Debt', which examined the conditions under which bank lenders will make concessions by taking equity in financially distressed firms. He developed a simple model to show that banks' role in debt restructuring depends whether the firm's capital structure includes public debt and whether it can be restructured. In his empirical work, based on a sample of 102 debt restructurings in the US during 1981-90, he noted that banks never make concessions to firms with public debt outstanding unless the public debt-holders also restructure their claims. When banks do take equity, they usually obtain substantial proportions of the firms' stock and maintain their positions for more than two years. Firms in which banks hold stock also perform significantly better after restructuring than those in which banks take no equity.

Colin Mayer (School of Management Studies, University of Oxford, and CEPR) wondered whether banks worry about taking controlling shares in equity or becoming actively involved in restructuring. He also suggested that the choice of security type and the scale of the write-down are probably both interrelated and endogenous.

Arnoud Boot (Universiteit van Amsterdam) opened a round table on `Regulation' by presenting `Financial System Architecture', written jointly with Anjan Thakor, which distinguished the roles of capital markets and banks in the allocation of corporate credit. In capital markets, the interaction of many agents' rational expectations determines equilibrium securities prices, which then feed back into firms' real decisions. The information in banks' credit contracts has no analogous impact on the real sector, but bank finance provides a superior resolution of moral hazard. In an adverse-selection framework, borrowers with observable qualities below a certain cut-off value therefore choose bank financing, while others obtain their funds from capital markets. A new financial system for example in a former centrally planned economy is likely to be dominated by commercial banks, which will engage in greater credit rationing as uncertainty over loan demand increases (and the share of borrowers accessing the capital market directly rises in anticipation of this). A rise in borrowers' overall observable quality causes investment banks to engage in more financial innovation. Universal banks engage in less credit rationing than commercial banks and less innovation than investment banks. Capital markets are thus likely to be better developed and better able to capture market share in economies with universal banks than in those in which commercial and investment banks are functionally separate.

In `Bank Capital Regulation in General Equilibrium', written jointly with Andrew Winton, Gary Gorton (University of Pennsylvania) investigated regulators' ability to use capital requirements to keep the banking system's stability at its (socially) optimal level. Gorton outlined a multi-period general equilibrium model in which banks issue debt claims to facilitate trading and invest in risky projects; this determines both the cost of bank capital and the banks' optimal capital requirements. Banks' unique ability to learn about projects, which is lost if they fail, also provides them with a valuable charter. Deposit insurance is socially valuable, but it also leads to moral hazard which requires adjustment of capital requirements. Regulators set initial capital requirements which they then adjust in the light of the riskiness of banks' portfolios. The regulators must take account of the private cost of bank capital which determines whether banks' equity-holders will agree to meet capital requirements rather than exit the industry. Gorton showed that the cost of banks' capital reflects their special role in the economy, constraints of private markets imply that capital markets are never binding, and regulators have a role to play in enforcing privately chosen equity levels if banks' charter values are low. Finally, even regulatory policies that are socially optimal in practice often resemble those resulting from a policy of `forbearance' regulators' selfish refusal to close banks with low or negative net worth to avoid becoming unpopular with bankers and/or politicians.

In `Competition versus Harmonization: An Overview of International Regulation of Financial Services', Lawrence J White (New York University) noted that the increasing globalization of financial services is both restricting the effectiveness of national regulation and leading national regulators to seek to harmonize key regulatory provisions by international agreement. Lawrence used the frameworks of market and government failure to assess the relative merits of competition and harmonization, distinguishing the cases of economic, safety and information regulation. He found that harmonization may enhance the efficiency of markets in financial services in a few specific instances: it may overcome government failure by inefficient (protectionist) national regulatory regimes, restrain national governments' tendency to subsidize domestic providers of financial services and reduce transaction costs. Policy-makers should focus attention on these cases, but competition both among national regulatory regimes and among firms nevertheless produces the preferred public policy outcome in most cases.

Russell Cooper (Boston University) then presented `Financial Intermediation and Aggregate Fluctuations: A Quantitative Analysis', written with Jo<176>o Ejarque, which investigated whether business cycle models in which aggregate fluctuations are driven by shocks to the process of financial intermediation can generate time series analogous to those observed in reality for the US economy. In his first model, fluctuations arise from disturbances in the intertemporal production process that creates capital tomorrow from investment today, interpreted as changes to the process that matches savings and investment. The model proved unsatisfactory since it yielded correlations among time series inconsistent with those observed in the data. In his second model, there were no shocks to tastes or technology and variations in the returns to intermediation formed part of a sunspot equilibrium. This model successfully reproduced the observed correlations among variables but yielded a variability in the capital stock that was unsatisfactorily large.

Ramón Caminal (Institut d'Anàlisi Econòmica, Barcelona) observed that these real business cycle models presented very crude characterizations of financial intermediation, omitting even such important variables as the spreads of lending and borrowing rates. Furthermore, in models with financial imperfections the equilibrium is often Pareto inefficient, so their predictions will not solve the planner's problem.

Anthony Santomero and Jeffrey Trester (University of Pennsylvania) then presented `The Effect of Asset Sales on Bank Risk in an Imperfect Information Environment', which developed an asymmetric imperfect information model to determine the overall effect on banks of reducing the costs of securitization and, in particular, whether the asset sales this induces will make their loan portfolios more or less risky. Securitizing assets enables a bank to convert illiquid instruments into liquid ones, and previous models of this type have suggested that banks will sell safe assets and keep risky ones, since they can signal the lower risk of a safe asset by using credit guarantees. The authors used their richer model of securitization to show that making securitization easier has an ambiguous impact: while the share of risky assets in banks' portfolios increases, their vulnerability to liquidity shocks is also reduced. Transforming an illiquid into a liquid asset may thus reduce the bank's overall exposure to risk.

Sugato Bhattacharya (Carnegie Mellon University) suggested reinterpreting this analysis in terms of substitution and income effects: easier securitization induces both a rise in the value of any portfolio (income effect) and a switch to a riskier asset mix (substitution effect).

In `A Bargaining Model of Financial Intermediation', Helmut Bester (Universiteit Tilburg and CEPR) considered the role of financial intermediaries as a commitment device for lenders. Bester considered the case of a lender who incurs high search costs in identifying a suitable investment project and must then engage in negotiations with the project's owner over how to split the resulting revenue, whose outcome may be predicted by a bargaining model with an outside option. Alternatively, search may be delegated to a financial institution which commits to pay the investor a given amount and must bargain with the project owner over the revenue split. The financial institution's commitment to pay the investor a fixed amount ensures that the outcome of this negotiation differs from that realized through direct bargaining. For some parameter values, an investor will be better off by delegating search to the financial institution, so intermediaries are basically used as a commitment device; intermediation can therefore arise even if financial intermediaries have no advantage in the search technology.

Sandro Brusco (Institut d'Anàlisi Econòmica, Barcelona, and CEPR) observed that an important feature of the model is the lack of a market for the search activity by lenders. A financial intermediary thus appears to serve as a means of overcoming problems related to the incompleteness of markets.

Bruno Biais (Université des Sciences Sociales, Toulouse) presented `Why do Firms Use Trade Credits? A Signalling Model', written with Christian Gollier and Pascale Viala. In their model, both `risky' and `safe' firms have private knowledge of their own riskiness, on which their suppliers also acquire information in the course of their business relationships. Suppliers are therefore well placed to assess creditworthiness and provide trade credit on terms reflecting their perceptions of that risk. Customer firms can use such credits to signal low riskiness credibly to other lenders including banks and obtain their loans under better conditions. The existence of three types of agent (customers, suppliers and banks) admits the possibility of collusion. In particular, the supplier may collude with a risky customer to create trade credit and thus obtain better terms from the bank. This is ruled out in equilibrium, however, by the incentive compatibility condition that the proportion of trade credit must be high enough for the cost of the signal to the risky firm to be larger than the surplus extracted from the bank.

Jan Bouckaert (Universiteit Tilburg) presented `Phonebanking', written with Hans Degryse, which investigated the effect of financial innovation on banking competition. In their model of a two-stage game, banks first choose whether to adopt a financial innovation that is valued by depositors (phonebanking) and later compete in the market for deposits. In the second stage, depositors incur transportation costs for certain transactions and therefore prefer to be closer to their banks, but they also value the option of phonebanking. Introducing this option makes a bank more attractive to customers but also increases competition among banks. In the two-stage game, different sub-game perfect equilibria may arise depending on the magnitude of the financial transaction types. Specialization is observed when phonebanking is introduced by only one bank.

Angel de la Fuente (Institut d'Anàlisi Econòmica, Barcelona, and CEPR) then presented `Innovation, <169>Bank<170> Monitoring and Endogenous Financial Development', with José Maria Marin, the first of three papers in a round table on `Finance and Growth'. He outlined a simple model of the interaction of capital accumulation, technological progress and financial development in a growing economy. Growth is sustained by the development of new products, which is a risky activity whose probability of success depends on the actions of entrepreneurs which are imperfectly observable and may be partially monitored at a cost. Financial intermediaries emerge to avoid duplication of monitoring activities, and both the degree of monitoring and the rate of innovation are endogenously determined in equilibrium. Monitoring is financial intermediaries' main activity, so the volume of resources absorbed by the financial sector and its efficiency in risk pooling are also determined endogenously. Closer monitoring improves risk sharing and thereby yields a higher level of innovative activity in equilibrium, which in turn raises productivity growth, income and capital accumulation. Under plausible assumptions, the resulting changes in factor prices reduce the costs of intermediation, which leads to further improvements in the financial sector's efficiency.

John H Boyd (Federal Reserve Bank of Minneapolis and University of Minnesota) presented `Capital Market Imperfections, International Credit Markets, and Nonconvergence', written with Bruce Smith, which investigated the impact of financial market liberalization on less developed economies. In his model of growth, capital investment requires some external finance and is complicated by the presence of a `costly state verification' problem, which may lead to credit rationing. For two economies whose initial levels of capital differ, their output levels will converge when both are closed to international capital lending, as they display monotonic adjustment towards the same steady-state level of the capital stock. If credit markets operate internationally, however, this convergence result is no longer guaranteed. While a country with a higher current capital stock has a lower marginal product of capital, its higher income implies a greater ability to finance capital investments internally, which mitigates the costly state verification problem. As a result, asymmetric equilibria may exist for which the two countries' capital stocks will not converge to the same level: investment funds will then flow from the poorer to the richer country.

Oren Sussman (Hebrew University of Jerusalem) presented the final paper, `Banking and Development', written with Joseph Zeira, which reformulated and supplied new empirical evidence to support the `financial development conjecture': that there are strong feedback effects exist between real and financial development. Sussman first outlined his information-based theoretical model of a monopolistically competitive banking system, in which financial intermediation provides a means of reducing the costs of monitoring the ex post returns to defaulting projects, and feedback between real and financial sectors reflects banks' specialization. Sussman focused on the cost of financial intermediation per unit of loan to formulate testable predictions in the empirical part of the paper. He used US cross-state aggregate data on banks are to show that the cost of banking is negatively correlated with the level of real economic development.