The Frisch elasticity of labor supply can be estimated by regressing hours worked on the hourly wage rate, controlling for consumption of the individual worker. However, most household panel surveys contain consumption information only at the household level. We show that proxying individual consumption by household consumption biases estimated Frisch elasticities downward as limited commitment in the household induces individual consumption to behave differently from household consumption. We develop an improved estimation approach that eliminates this bias by exploiting information on the composition of household consumption to infer its distribution. Using PSID data, we estimate Frisch elasticities of about 0.7.