This paper analyses the effects of an introduction of a retail central bank digital currency (CBDC) on bank intermediation in a tractable general equilibrium model with heterogeneous bank deposits and an imperfectly competitive loan market. The agents in the economy have preferences over holding central bank money or bank deposits that are not solely based on the assets' returns. I find that the impacts of a CBDC strongly differ depending on whether it is used only as a payment vehicle that competes with short-term transaction deposits or also as a saving vehicle that competes with longerterm saving deposits. A calibration of the model to the US economy from 1987-2006 shows that if a CBDC is only used as a payment vehicle, a 10% outflow of agents from transaction deposits to CBDC decreases bank lending by 1.2%-1.3%. The effect is almost three times stronger if a CBDC is also used as a saving vehicle in which case bank lending shrinks by about 3.0%-3.3%.