A CEPR workshop on ‘The Location of Financial Activities’ was held in Naples on 17/18 May. It was organized by Thomas Gehrig (Universität Basel and CEPR), Marco Pagano (Università di Napoli Federico II, CNR–ISFSE, and CEPR) and Jacques Thisse (Université Catholique de Louvain and CERAS–ENPC, Paris, and CEPR). The workshop forms part of CEPR’s research programme on ‘The New Economic Geography of Europe: Market Integration, Regional Convergence and the Location of Economic Activity’, supported by the European Commission’s Human Capital and Mobility programme. Additional support was provided by CNR–ISFSE (Consiglio Nazionale delle Ricerche-Istituto di Studi sulle Strutture Finanziarie e lo Sviluppo Economico).

Ingrid Werner (Stanford University) presented ‘Effects of Geography and Stock Market Structure: A Comparison of Cross-Listed Securities’, written with Allan Kleidon (Cornerstone Research and Stanford University). This paper examines international intra-day data for British securities traded in multiple markets. In order to distinguish between the effects of market microstructure (the structure aspect) and location (the international aspect), four different markets are analysed: the London Stock Exchange (LSE), the NYSE, AMEX, and NASDAQ. The paper shows that both location and structure affect quote variance, transaction variance, volume and spread. The geographical effects reflect cultural behaviour (for example, trading volume in the UK is bimodal with a clear lull at midday) and hence do not affect security pricing. On the contrary, differences in market structures may explain the differences in security prices. The most striking example is the spread pattern around opening for NASDAQ versus NYSE/AMEX stocks. The authors find that spreads rise sharply on NASDAQ immediately after opening, while for the other markets a decline in spreads is observed. The paper concludes that theoretical models need to incorporate more descriptive realism to increase the understanding of different types of markets.

Bernard Dumas (Hautes Études Commerciales, Jouy-en-Josas, and CEPR) questioned whether it was useful to relate the differences in spread behaviour to those in volume and volatility behaviour, since all of these are endogenous variables. He suggested including instead some exogenous variables. Pagano was surprised by the result of declining spreads throughout the day in the dealer markets, emphasizing that the typical intra-day pattern of quoted spreads for stocks cross-listed on the London market and European markets is U-shaped. Oved Yosha (Tel-Aviv University) raised the problems of arbitrage across markets and of the home bias in portfolio choice, which were not considered in the paper.

Laura Bottazzi (IGIER, Università Bocconi, Milano, and CEPR) presented ‘Wages, Profits and the International Portfolio Puzzle’, written with Paolo Pesenti (Princeton University) and Eric van Wincoop (Boston University). Their paper investigates the impact of fluctuations in the return to human capital on the composition of international asset portfolios. The lack of diversification across countries in portfolio choice is in contrast to the theoretical paradigm of global diversification strategies. The authors assert that this discrepancy can be explained by the role of human capital. Shocks that lead to a redistribution of income between capital and labour, create a negative correlation between the return to human capital and the return on domestic equities, making foreign securities a less attractive hedge against labour income risk. To test this hypothesis, the authors use a continuous-time VAR model of international portfolio choice, which allows for intertemporal interactions between wage rates and capital returns. The data set used contains information on sixteen countries. The evidence accounts for an average home bias of 30–35% in favour of domestic securities, although there are great differences across countries, and few exceptions to the rule. This leads the authors to conclude that the presence of fluctuations in non-tradable labour incomes helps rationalize the direction of the portfolio bias, even though it does not lead to a complete ‘solution’ of the puzzle.

Giovanna Nicodano (Università di Torino) questioned the relevance of trying to estimate the distribution of financial assets among wage owners and entrepreneurs. She argued that in the current version of the work the home bias has been overestimated. Dumas suggested examining the composition of the whole market portfolio to clarify some differences among countries and to consider the issue of the size of countries. It was pointed out that many households do not own stocks. Bottazzi replied that many do hold stocks indirectly via pension funds, but Werner noted that most pension funds are mainly composed of bonds, and therefore government bonds should be included in the analysis.

‘The Total Cost of Trading Belgian Shares: Brussels versus London’ was presented by Hans Degryse (Université de Lausanne and CEPR). This paper investigates competition between the quote-driven SEAQ International (SEAQ-I) market and the order-driven Brussels CATS system, by using a simultaneous record of quotes, limit orders, and transaction prices for Belgian shares. The data set contains the entire limit order book and also ‘hidden’ orders. The results of the paper confirm that on SEAQ-I small and medium trades are more expensive than on CATS: the Brussels market is considerably tighter than SEAQ-I, while the latter is deeper. There is evidence that the relation between effective spread and trade size is U-shaped on the Brussels market. Conversely, there is little evidence of any trade size effect on SEAQ-I for Belgian equities.

Ailsa Röell (ECARE, Université Libre de Bruxelles, and CEPR) noted that the paper is valuable in giving accurate estimates of the liquidity available in the limit order book for reasonable trade sizes. She was surprised by the strong evidence of U-shaped effective spread in Brussels, which implies the presence of non-neglible costs in the auction market, since in the auction fixed costs are expected to be lower than in the London dealer market.

‘Is There Scope for Fragmentation in Financial Markets?’ is the question Konrad Stahl (Universität Mannheim and CEPR), Thomas Gehrig and Xavier Vives (Institut d'Anàlisi Econòmica (CSIC), Barcelona, and CEPR) try to answer. They investigate under what conditions fragmented markets survive an exogenous integration process. An important assumption of their model is that local investors enjoy a comparative informational advantage over non-local ones, because information about firms is not ubiquitous but concentrated in each firm’s location. The relevant results of the paper are the following: first, trade in fragmented stock exchanges may constitute an equilibrium outcome when private information is valuable and market access costs for investors are large enough. Second, in the absence of asymmetric information, concentration of listings in a single exchange is always an equilibrium outcome when investors are few. Nevertheless, it will not necessarily be Pareto preferred. Lastly, the introduction of internationally active firms, whose activities are publicly observed, may or may not upset a fragmentation equilibrium, depending on whether cross listing is possible.

Patrick Bolton (ECARE, Université Libre de Bruxelles, and CEPR) suggested empirically estimating the magnitude of the effects in the model. He noted that the model does not constitute a General Equilibrium Model, since there are no endowments for investors, and it is not clear where the proceeds of the firms go. Another remark concerned the assumption of risk neutrality for firm managers which is not consistent with the other assumptions of the model. Dumas suggested considering a firm-specific market access cost for investors, and assuming it to be related to the size of the firm.

The ‘Financial Development Conjecture’ – namely the idea that there are strong feedback effects between real economic development and financial development - is reformulated in the paper ‘Banking and Development’, presented by Oren Sussman (Hebrew University of Jerusalem) and written with Joseph Zeira (Hebrew University of Jerusalem and CEPR). Their theoretical model explains the relation between real and financial development through banks’ specialization and the cost of financial intermediation. The two main assumptions are that financial intermediation is a costly activity based on monitoring ex-post returns of defaulting projects. These costs are increasing with the distance between the bank and the project and declining with specialization. In this context the feedback effects, between real and financial development, work as follows. As the economy grows, more capital is channelled through the banks, which raises their profits and induces the entry of more banks. This entry reduces the average distance between banks and projects, increases specialization and thus reduces the cost of financial intermediation. This, in turn, raises investments and real growth. In the empirical part of the paper, US cross-state data from banks’ income statements are used to measure the correlation between real economic development and the cost of financial intermediation. The estimation results are consistent with the theoretical model and show that the cost of banking is negatively related to the level of real economic development.

Hans Gersbach (Universität Heidelberg) stated that in the model there is no labour component in the banks’ set-up costs. If this component is introduced, then the entry cost of banks would increase with economic development and there would be a counter effect to the specialization effect, which could reverse the results. Gersbach also claimed that in the model banks monitor only defaulting projects while in reality they monitor primarily non-defaulting projects. He suggested introducing random monitoring in the paper and check how this would affect the specification of the regression. Lastly, he suggested that the model should be interpreted as a branch model because in reality more developed countries have fewer banks. Pierre-André Chiappori (DELTA, Paris, and CEPR) commented that using the total cost of banking as an approximation of the monitoring costs is not very convincing since the latter is just 10% of the former – he suggested estimating the correlation between the monitoring cost and the business cycle.

In ‘Excessive Risks and Banking Regulation’ Thomas Gehrig analyses the incidence of regional risk in a spatial model where banks specialize in serving a particular market segment and compete for deposits and loans. Entry of new banks in the market is socially beneficial to the extent that it reduces transaction costs for depositors and borrowers, but it is also socially costly since it requires sunk investments. The paper shows that entry can be either excessive or insufficient relative to the welfare-maximizing industrial structure depending on the relative mobility of borrowers and lenders. Moreover, with free entry insolvency rates of banks are high relative to the social optimum because they are negatively related to the concentration of the banking industry. This suggests that the regulation of banks should be guided by the notion that competition among banks may be excessive, exposing them to considerable risk of failure. For example, restrictions on entry can improve social welfare and reduce insolvency risk because the oligopoly profits of banks may be used as buffers against regional shocks. Lastly, the analysis in the paper suggests that the imposition of capital adequacy rules may increase the probability of bank failures if they are not complemented by additional entry restrictions.

Pierre-André Chiappori noted that the real competition between banks takes place in the borrowing market since, once banks attract depositors, they are capacity-constrained in the lending market. In this context, the competition in the borrowing market pushes up the borrowing rates offered by the banks but, as a consequence, also the lending rates. Therefore, without a money market, both borrowing and lending rates are more likely to be high. He argued that the paper does not consider any social cost of banks’ failure and so the optimal number of banks is always overestimated.

The paper ‘Monetary Union or Else’, presented by Hans Haller (Virginia Polytechnic Institute) written with Yannis Ioannides (Tufts University), analyses economic integration in a pure exchange economy as a two-stage strategic game involving two countries. In the first stage, each country chooses simultaneously and independently whether to keep its own currency or to adopt the other country’s currency. If both countries agree on which currency to adopt, then a monetary union emerges. In the second stage, the two countries commit resources to economic integration simultaneously and independently. Only if the total investment by the two countries is sufficiently high, is economic integration achieved. A common currency reduces the overall cost of economic integration but imposes an idiosyncratic adjustment cost on the country changing its currency. Therefore, a country is willing to change its currency in the first stage only if it expects the other one to contribute more to the cost of economic integration in the second stage.

Marco Da Rin (IGIER, Università Bocconi, Milano) argued that the paper raises more questions and more issues than it actually solves, and that the timing structure assumed in the paper is at odds with the European experience. He suggested an inversion of the timing structure (first the real, then the monetary integration) to analyse if and how it affects the results of the paper. In this case, the two countries would face a trade-off between the loss of exchange rate flexibility and seignorage and the reduction of transaction costs. He also suggested considering a third currency in the model so that the countries have another possible choice.

Oved Yosha (Tel-Aviv University) presented ‘Channels of Interstate Risk Sharing: United States, 1963–90’, a paper written with Pierfederico Asdrubali and Bent Sorensen (Brown University). There are three channels through which risk sharing can occur in a federal regime. First, the members of the federation can share risk via cross-ownership of productive assets (capital market smoothing). Second, the federation’s central government can smooth income through the tax-transfer system (federal government smoothing). Third, the members of the federation can smooth their consumption by adjusting their asset portfolio (credit market smoothing). The paper shows that, for the period 1963–90, 39% of the shocks to gross state product were smoothed by capital markets, 13% by the federal government and 23% by credit markets. The remaining 25% were not smoothed. The hypothesis of full interstate risk sharing is, therefore, rejected. Moreover, the paper states that in recent years an increase in capital market and federal government smoothing has been observed, whereas credit market smoothing has been less stable over time. Lastly, the paper decomposes the federal government smoothing into sub-categories. In particular, it is shown that the federal tax system smoothes 4.3% of changes to gross state product, the transfer system 6.3%, unemployment benefits 1.9% and grants to states 2.5%.

Tullio Jappelli (Università di Salerno and CEPR) pointed out that the conclusions of the paper would be stronger if the shocks to gross state product are idiosyncratic. He was surprised by some of the empirical results, in particular, he was puzzled by the fact that unemployment benefits explain less than 2% of consumption smoothing while capital market smoothing is quite large, although in the United States only 25% of households hold any form of financial assets. Moreover, he pointed out that the result of recent decline in total smoothing is at odds with the evidence on financial innovation and the lifting of state branching regulation.