We consider an overlapping generations economy where capital is produced from bank loans under stochastic constant returns to scale, and subject to idiosyncratic shocks whose realisations are costly to verify. Our formulation differs from earlier work in permitting investment projects to be infinitely divisible and private agency costs to be convex. If there are external economies to financial intermediation, then deviations from steady-state output are negatively correlated with the spread between loan and deposit rates. Moreover, the capital stock correspondence is set-valued, a result consistent with poverty traps, growth cycles, and humpshaped impulse response functions. This paper investigates how nonconvexities in financial intermediation, particularly in the cost of collecting private information about borrowers, enrich the operating characteristics of a simple, and relatively tractable, onesector growth model. The enriched set of equilibria is consistent with a number of phenomena that are hard to reconcile with convex models of economic growth: poverty traps, rank reversals, growth cycles and humpshaped impulse response functions are all possible outcomes of nonconvex information costs. Bankers and economists alike have long regarded the credit market as key to understanding economic development and to transmitting cyclical shocks through modern industrial economies. The growth branch of this literature starts with Gurley and Shaw (1967) who note that economic growth is almost universally accompanied by financial deepening, that is, by more intensive use of external finance in investment and by a gradual lifting of distortions in the credit market. The cyclical fluctuations branch of the literature focuses on the connection between credit market conditions and business cycles; a key concern here is how the credit market propagates and amplifies external shocks through the entire economy. The general idea dates back at least to Keynes, Fisher, and Friedman and Schwartz who argued that adverse conditions in financial markets may have exacerbated the effects of prewar recessions, including the Great Depression. Greenwood and Smith (1995) survey work on development. Among recent contributions, we note Bencivenga and Smith (1991) who study the growth effects of financial intermediation in an overlapping generations model with uncertain liquidity needs. Intermediation enhances growth because banks are efficient providers of liquidity which frees individuals from the need to maintain low yielding liquid assets. Similar results are obtained in Greenwood and Jovanovic (1990), where intermediary institutions are shown to arise