In the data, we observe persistent loan rates dispersion that violates the law of one price for homogeneous loan products. Moreover, loan rate dispersion is positively correlated with average loan markups by financial institutions and banks. Details matter: In order to design good public policy, one needs a policy-analysis framework where explicit liquidity and medium of exchange roles of money and credit are equilibrium phenomena. It has been shown that under a perfectly competitive banking system, banks unambiguously improve welfare. This is because banks help provide an insurance role against agent’s idiosyncratic risk of holding idle liquidity (Berentsen, Camera and Waller, J. Econ. Theory, 2007). We generalize this theory by modelling noisy consumer search for loans. (Consumer search models are also popular in the Industrial Organization and Marketing literature.) As a consequence, there is an equilibrium and policy-dependent market structure with imperfect competition in the loans market. Positively, our equilibrium model rationalizes empirical observations on the relationship between loan rate dispersion and markups. Normatively, we show why central banks and regulators are rightfully concerned with banks having too much market power: In our setting banks trade-off their intensive and extensive margins of loan sales and as a result, depending on policy and economic primitives, the basic welfare insurance role of banks gets eroded by equilibrium market power in that industry.