This paper investigates a new mechanism through which "liquidity effects" (i.e. a negative response of the nominal interest rate to monetary injections) can be introduced into dynamic stochastic general equilibrium (DSGE) models. As it turns out, this liquidity effect has been found difficult to obtain in standard monetary DSGE models. The reason is an "inflationary expectations effect": when an unexpected money injection occurs, this creates the expectation of further money increases in the future, which itself creates the expectation of future inflation. From Fisher's equation this will tend, ceteris paribus, to raise the nominal interest rate. Inspite of this effect, one can find in the literature a few models and mechanisms which introduce a liquidity effect in DSGE models. Two prominent ones are: (1) Models of limited participation where households cannot adapt immediately their financial portfolios when a monetary policy shock occurs. (2) Models of sticky prices, where prices are preset in advance. In this paper we explore a third avenue, which we call "non-Ricardian". By non-Ricardian we mean models, like the overlapping generations model of Samuelson (1958), where new agents are born every period, and where "Ricardian equivalence" does not hold. We argue that the non-Ricardian character of these economies is conducive to a liquidity effect. Indeed in Ricardian economies the value of financial assets is matched by an equivalent value of future discounted taxes, so that they are not real wealth to the agents. Instead in non-Ricardian economies part of the agent's financial wealth represents real wealth to them, because they can pass it to later generations against real goods, and the next generations will bear part of the tax burden. As a result this creates an effect similar to Patinkin's (1956) famed "real balance effect". This effect itself gives rise to the liquidity effect. In order to make this intuition more formal, we use a model adapted from that of Weil (1987, 1991) which "nests" the usual Ricardian model with an infinitely lived consumer. Households never die, but new "generations" are born each period at a rate n. After long calculations we find an equation giving the interest rate as the sum of two terms: the first one displays the "inflationary expectations" effect. The second one gives the liquidity effect described above. That second effect is present only if n>0, i.e. if the economy is actually non-Ricardian, and it is greater the higher n is. We give sufficient condition for this liquidity effect to overrule the inflationary expectations effect, and to be persistent in tim