The ?prospect of upward mobility? (POUM) hypothesis formalised by Benabou and Ok (2001a) finds explicit assumptions under which some individuals that are poorer than the average optimally choose to oppose redistribution policies. The underlying intuition is that these individuals rationally expect to be richer than average in the future. This result holds provided the mobility process is concave in expectations, redistribution policies are expected to last for a sufficiently long period and individuals are not too risk averse. This paper tests the POUM hypothesis by means of a within subjects experiment where the concavity of the mobility process, the degree of social mobility, the knowledge of personal income and the degree of inequality are used as treatments. Other determinants of the demand for redistribution, such as risk aversion and inequality aversion are (partially) controlled for via either the experiment design or the information collected during the experiment. We find that the POUM hypothesis holds under alternative specifications, even when we control for individual fixed effects.