Introduction
That financial structure and economic development are interrelated is a well known hypothesis (see Goldsmith, 1969, McKinnon, 1973, Shaw, 1973, Kuznets, 1971, Cameron, 1967 and Townsend, 1983; see also Gertler, 1988, for an excellent survey of the background and Greenwood and Jovanovic, 1989, for a recent contribution). In the present paper, we utilize some recent developments in the theory of contracts and the organization of financial markets (most notably Diamond, 1984, and Gale and Hellwig, 1985), in order to reformulate the financial development hypothesis in a way which is both theoretically comprehensible, and empirically testable. We focus on the ‘gross markup’ of the banking system, roughly the gap between the (real) rate at which banks borrow and lend money. This gap reflects (in addition to the more conventional default-premium component), the cost of financial intermediation. In the theoretical part we construct a spatial model of a monopolistically competitive banking system. When the capital stock increases, the market for financial intermediation grows, and the number of banks increases (due to entry). Each bank becomes more specialized, and thus efficient, over a smaller market share. Also, the industry becomes more competitive. As a result the cost of intermediation falls and the markup decreases. In the empirical part we present some evidence showing that the markup is lower in high-income countries relative to low-income countries.