A joint workshop with the Hebrew University of Jerusalem (Pinhas?) on `Financial Intermediation and Real Economic Activity' was held in Jerusalem on 7/8 May. The workshop was organized by Jean-Pierre Danthine (Université de Lausanne and CEPR), Philippe Weil (Université Libre de Bruxelles and CEPR) and Joseph Zeira (Hebrew University of Jerusalem and CEPR). Additional financial support was provided by the Hebrew University of Jerusalem, Bank of Israel, Bank Leumi, the Maurice Falk Institute of Economic Research, the Pinhas Sapir Economic Policy Forum and the Tel Aviv Stock Exchange.

The first paper was `Competition, Financial Discipline and Growth' by Philippe Aghion (Nuffield College, Oxford, EBRD and CEPR), Mathias Dewatripont (ECARE, Université Libre de Bruxelles and CEPR) and Patrick Rey (ENSAE-CREST, Paris and CEPR). The paper introduces agency problems into a Schumpeterian growth model, focusing on technology adoption by individual firms, and adding a learning externality whereby growth depends positively on the rate of technology adoption. In this model, product market competition contributes negatively to growth when firms are run by profit-maximizing managers. This Schumpeterian market is overturned when firms are run by `conservative' managers, determined to delay technology adoption as much as possible subject to keeping the firm in the market. Then, competition fosters growth by reducing managerial slack, since they enjoy lower net cash inflows, and industrial policy in the form of adoption subsidies tends to reduce growth. The paper then introduces financial discipline in the form of debt contracts that commit firms to pay out cash, thereby reducing slack. In this world, product market discipline and financial market discipline are substitutes instead of complementary instruments to reduce slack and foster growth.

Zeira commented that this paper belongs to the new literature on growth convergence. It examines an additional explanation to differences between countries, the degree of competition. The reaction of growth to competition depends on the assumption about managerial types, but these strong results depend to a large extent on the way competition is modelled. If markets for intermediate goods are not monopolistic, competition can raise growth even when managers are profit maximizers.

In `Fiscal Policy, Productivity Shocks and the US Trade Balance Deficit' by Robert Kollman (Université de Grenoble and Université de Montréal), a two-country real business cycle model is used to study the behaviour of the US trade balance. Economic fluctuations are driven by productivity shocks and by variations in government purchases and distorting taxes. The model is simulated using quarterly data on total factor productivity government purchases and the average tax rate in the seven major industrial countries between 1975-91. A version of the model that postulates complete international asset markets, as frequently assumed in the literature, fails to explain the observed behaviour of the US trade balance. In contrast, a version with incomplete asset markets, in which only debt contracts can be used for international capital flows, tracks the behaviour of the US trade balance fairly closely.

Fabio Canova (Universitat Pompeu Fabra and CEPR) commented that the model explains the size and cyclical variations of US trade balance under incomplete asset market because it implies a very close relationship between the trade balance and import of final goods. The data seem to indicate that the deficit of the 1980s was primarily concentrated in intermediate industrial and consumer durable goods. The model implies a trade surplus for investment goods, which are the closest proxies to these quantities in the model, both under complete and incomplete markets, and so it cannot really explain the causes of the trade deficit. Also, simulated consumption correlations are negative in the model with incomplete markets, which is never the case in international data.

In `Do Capital Market Imperfections Exacerbate Output Fluctuations?', Philippe Bacchetta (Studienzentrum Gerzensee and Université de Lausanne) and Ramon Caminal (Institut d'Anàlisi Econòmica and Universitat Autònoma de Barcelona) develop a dynamic general equilibrium macroeconomic model where a proportion of firms are credit constrained due to asymmetric information. In general, a macroeconomic shock has additional effects created by a reallocation of funds between credit constrained and other firms. The paper shows, however, that the output response to shocks is not necessarily amplified and can be dampened by the presence of asymmetric information. This depends on the impact of the shock on the composition of external and internal funds for firms. Furthermore, it is important to distinguish between firms' collateral and cash flow in determining the dampening or amplifying effect of agency costs.

Oren Sussman (CEMFI, Madrid) commented that while capital market imperfections may amplify the effect of exogenous shocks on output fluctuations, the opposite may happen too. The result has important policy implications: if the government's goal is to stabilize the economy, the direction of that intervention is not clear in absence of robust comparative statics results. Hence, any argument in favour of intervention based on capital market imperfections is bound to boil down to a fine tuning argument. Steven Durlauf (University of Wisconsin at Madison) noted the paper's two contributions to the literature on credit constraints and aggregate activity. First, it develops a general equilibrium model allowing for competition between firms which are and are not credit constrained for available funds. Second, it shows that this competition implies that there is no unambiguous link between capital market imperfections and the degree of output `sensitivity' to various types of innovations. Intuitively, when firms with different marginal products of capital compete for funds, substitution between them can dampen fluctuations. This is very different from models driven exclusively by changes in the internal funds of the constrained firms.

`A Macroeconomic Model with Imperfect Competition in Banking' by Rafael Repullo (CEMFI and CEPR) constructs a fully consistent general equilibrium model of a monetary economy in which banks have market power in relation to both depositors and borrowers. This is used to analyse how changes in the degree of competition, the minimum reserve requirement or in the rate of growth of the monetary base (the rate of inflation), or the introduction of a ceiling on deposit interest rates, affect the steady state equilibrium and the welfare of agents. It is shown that a reserve requirement is only equivalent to a tax on deposit rates in the limit case of perfect competition.

Danthine commented that the model is an elegant and sophisticated construct, but somewhat in search of an issue to address. The dimensions along which banks' market power appears to matter should be stressed, including its role (relative to a model with costly intermediation) in the non-equivalence between reserve requirements and tax on deposits. Fabrizio Zilibotti (Universitat Pompeu Fabra) suggested possible extensions in the direction of: first, providing a `rationale' for the existence of financial intermediaries; and second, modelling entry and allowing for endogenously variable mark-ups which change both throughout the business cycle and as a function of the development level. Some comparative static exercises show that many results of standard competitive models carry over to the micro-founded framework presented. The second extension could reveal that the two models perform differently in terms of some dynamic predictions, for example, how the behaviour of the banking system amplifies productivity shocks.

In `Why and to What Extent Do Investment Euler Equations Fail?', Toni Whited (University of Delaware) attempts to understand the reasons for the extent of the well documented failure of empirical investment Euler equations. The paper stays within the internally consistent maximizing framework of the null model and the two new econometric techniques. First, it estimates a signal to noise ratio as a metric for model performance. The null model with linear marginal adjustment costs fails all specification tests, while adding non-linearities improves the model to some extent. Second, the paper tests the model's overidentifying restrictions separately and finds that financial variables contribute to its failure. The most important culprit, however, is the misspecification of the marginal product of capital.

Leonardo Leiderman (Tel Aviv University and CEPR) suggested potentially interesting extensions. First, dynamic simulations based on the Euler equations could be performed to see how they fit the data. Second, it would be useful to specify and test more detailed alternatives to the null hypothesis, which incorporate a more explicit modelling of potential effects of financial structure and liquidity constraints. Third, it may be useful to test the restrictions of a more expanded view of the investment decision, taking account of other firm decisions regarding demand for variable factors. Fourth, the empirical implications of Dixit-Pindyck type models of irreversible investment could be explored. Bent Sorensen (Brown University and University of Copenhagen) noted that the paper extends the usual quadratic adjustment cost function to higher order terms. Furthermore, it uses information about the `fit' of the equation using real financial variables in order to obtain indications of potential sources of misspecification.

`Combining Arm's-Length and Control-Oriented Finance: Guidance from Main Bank Relationships in Sweden' by Erik Berglöf (Université Libre de Bruxelles) and Hans Sjogren (Stockholm University) examines bank-firm relationships in Sweden between 1985-93. The paper shows that most companies combine control-oriented financing through a dominant single bank lender with an arm's-length relationship to a second bank. The main bank utilizes more of its credit limit, has level collateral and holds more equity through related institutions than the side banks. Turnover of main bank relationships has also increased, suggesting that the role of the bank has shifted from monitoring to providing ex post flexibility to firms in liquidity difficulties.

Colin Mayer (University of Oxford and CEPR) noted that the main issues raised by the paper are: first, the definition of the main bank (its definition based on the relative size of loans extended by banks does not accord well with the common interpretation of a main bank); second, the statistical significance and robustness of the results compared to more formal econometric modelling; and third, the precise influence of the Swedish banking crisis on the reported results. The crisis offers the opportunity to investigate how a shock to the financial system affects bank-firm relations: do some companies suffer more than others in terms of their access to finance, which companies switch banks, and do banks take more or less equity shares? Yishay Yafeh (Hebrew University of Jerusalem) suggested that the paper should combine more financial information on firms in the sample: what are the attributes of main bank clients? What characterizes firms with partial bank ownership? Do banks hold equity in firms to foster `relationships'? Why do firms attempt to contest the monopoly power of their main bank by approaching a second bank rather than an arm's length capital market? Why is financial liberalization often accompanied by a decline in bank-firm ties?

`Channels of Interstate Risk-sharing: US 1963-90' by Pierfederico Asdrubali (Brown University), Bent Sorensen and Oved Yosha (Brown University and Tel Aviv University) develops a framework for quantifying the amount of risk-sharing among US states, and construct data which allows the decomposition of a shock to gross state product into several components. The paper finds that 40 per cent of shocks to state `gross domestic product' are smoothed by capital markets, 14 per cent are smoothed by the federal government, and 24 per cent by credit markets. The remaining 22 per cent are not smoothed. The federal government smoothing is decomposed into sub-categories (taxes, transfers, and grants to states), indicating, for example, that in comparison to the tax-transfer system, the magnitude of smoothing through the grant system is small (2.7 per cent of a shock), and that the unemployment insurance system smoothes 1.8 per cent of a shock. Analysis of two sub-periods shows that the amount and composition of federal government smoothing is stable through time, but there is an increase in the amount of capital market smoothing, a sharp decrease in the amount of credit market smoothing, and a decrease in the overall fraction of a shock smoothed.

Caminal thought that the paper provides interesting stylized facts about different channels of US consumption smoothing, but that the absence of an underlying structural model makes interpretation difficult, and any policy implications impossible to draw. Francesco Giavazzi (IGIER, Universita Bocconi and CEPR) noted that the paper makes the obvious, but important, point that fiscal transfers and financial integration are substitutes. This is relevant in connection with the debate on EMU. He noted, however, that the paper overlooks the role of inter-state migration, a potentially important channel of US risk-sharing, even at the yearly frequency considered.

In `Transition and Financial Collapse', Harald Uhlig (Tilburg University) shows that a collapse of the entrepreneurial sector in transition countries is a theoretical possibility due to the low stock of experienced entrepreneurs in addition to the low stock of initial productive capital. Remedies are discussed; the most promising are creating inequality and massive long-term subsidies.

Daniel Tsiddon (Hebrew University of Jerusalem and CEPR) noted that the paper presents a general equilibrium model where financial markets suffer from both moral hazard and adverse selection. It shows that financial collapse may occur in an economy where at there is not enough private capital in the hands of entrepreneurs to allow for loans at reasonable prices. As a result of collapse, the economy converges to a low level of economic activity. While the model reinforces similar results in other models, applying it to Eastern Europe is not altogether convincing. Specifically, the collapse does not change relative prices (it is a one sector model) and there is no way to accumulate capital over time. Ernst-Ludwig von Thadden (Universitat Basel) commented that the paper synthesizes several strands of the literature on development traps, by integrating production externalities, capital markets imperfections, and human capital acquisition. The resulting development dynamics are rich, but do not go beyond those of existing, simpler models.