Structure and performance in European banking
Two competing hypotheses with regard to market structure and performance are the traditional structure-conduct-performance (SCP) paradigm and the efficiency hypothesis. This thesis presents results for tests of both hypotheses with respect to the European banking industry using pooled and annual data for the period 1986 to 1989. The cross-sectional and pooled results generally support the traditional SCP paradigm as an explanation for the market behaviour of European banks, with little evidence to suggest that the efficiency hypothesis holds. We also find that changes in market demand conditions, the equity-to-assets ratio and the staff expenses ratio appear to be significant and positively related to banking industry performance. In the majority of cases the loans-toassets ratio exerts a negative influence on banks' profitability. The individual country estimates find evidence that the SCP paradigm unambiguously seems to hold in Belgium, France, Italy, the Netherlands and Spain. These findings are in line with the Price Waterhouse/Cecchini study which identified the same countries, apart from the Netherlands, as the markets which would experience the largest financial service price falls post 1992. As such, these banking markets appear to offer the greatest incentive for new entrants to benefit from (and compete away) high average industry margins. The second part of this thesis adopts a methodology which allows us to test for inter-firm behaviour between leading banks across European bankir- *markets. The initial findings indicate that a large leading bank appears on average to promote cooperation with other leaders and this, on average, increases banking industry profitability. A large second bank, however, seems on average to induce rivalry with leaders rather than cooperation. The impact of more distant rivals does not seem to affect the profitability of banks in the industry. Further investigation of these results however, reveals that there are estimation problems brought about by the way in which the interactive market share variables - which test for cooperation and rivalry - are calculated. The nature in u,iich these variables are constructed implies a collinearity bias when the market shares of the largest firms are of a similar size. As a result, the Kwoka and Ravenscroft (1986) approach adopted to test for cooperative and rivalrous behaviour in European banking may be inappropriate. If we re-specify the model, however, to take account of this collinearity problem, we still observe evidence of duopoly behaviour in European banking thus confirming our earlier findings.