Varian (1988) showed that the utility maximization hypothesis cannot be falsified when only a subset of goods is observed. We show that this result does not hold under the assumptions that unobserved prices and expenditures remain constant. These assumptions are naturally satisfied in laboratory settings where the world outside the lab remains unchanged during the experiment. Hence for so-called induced budget experiments the Generalized Axiom of Revealed Preference is a necessary and sufficient condition for utility maximization in general, not just in the lab. Lab experiments are therefore a valid tool to put the utility maximization hypothesis to the test.