We use a large, randomized field experiment that exogenously varies prices to test their effect on consumption. Full information is available at zero costs. However, households state prices that are, on average, ten times larger than actual. But ignorant households cannot react to prices and so the price mechanism becomes dysfunctional. Our results explain small or zero price effects from previous research. We show that researchers must provide evidence for a functional price mechanism before ascribing causal effects or risk biased conclusions. The same applies to price instruments that are often regarded as first best solutions.