Many empirical studies on intertemporal choice report preference reversals in the sense that a preference between a small reward to be received soon and a larger reward to be received later reverses as both rewards are equally delayed. Such preference reversals are commonly interpreted as contradicting constant discounting. This interpretation is correct only if baseline consumption to which the outcomes are added, remains constant over time. The difficulty with measuring discounting when baseline consumption changes over time, is that delaying an outcome has two effects: (1) due to the change in baseline consumption, it changes the utility from receiving the outcome, and (2) it changes the factor by which this utility is discounted. In this paper we propose a way to disentangle the two effects, which allows us to draw conclusions about discounting even when baseline consumption changes over time.