We propose a theory of payments that highlights a conflict between the roles of medium of exchange and store of value. We posit that payments must involve the reciprocal transfer of a scarce reserve good, which holds value for other non-payment purposes. The theory demonstrates that agents make payments only when reserves are abundant enough and when the conflict is low. Otherwise, history-dependent equilibria arise in which an agent’s payment decision depends on the payment history of other agents within an equilibrium. The theory explains why payments frequently encounter delays and interruptions. Improving payment technologies may not reduce such fragility when reserves remain scarce and valuable for non-payment functions. The theory helps explain the evolution of money and payment systems, encompassing metallic payments before fiat money, modern bank payments, cross-border payments, and contemporary digital payment systems.

The deposit business differs at large versus small banks. We provide a parsimonious model and extensive empirical evidence supporting the idea that much of the variation in deposit-pricing behavior between large and small banks reflects differences in preferences and technologies. Large banks offer superior liquidity services but lower deposit rates, and locate where customers value their services. In addition to receiving a lower level of deposit rates on average, customers of large banks exhibit lower demand elasticities with respect to deposit rate spreads. As a result, despite the fact that the locations of large-bank branches have demographics typically associated with greater financial sophistication, large-bank customers earn lower average deposit rates. Our explanation for deposit pricing behavior challenges the idea that deposit pricing is mainly driven by pricing power derived from the large observed degree of concentration in the banking industry.

We document that the provision of liquidity insurance by banks to other financial institutions poses a major system risk to the banking sector. Using credit lines to Real Estate Investment Trusts (REITs) as a laboratory, we find that drawdown rates of REITs are higher than the ones of non-financial borrowers and more sensitive to aggregate market stress. This translates into higher tail risk and lower stock market returns for banks more exposed to REITs. Surprisingly, banks do not price these risks and offer cheaper credit lines to REITs than to other borrowers.