In Medicare Part D, low income individuals receive subsidies to enroll into insurance plans. This paper studies how premiums are distorted by the combined effects of this subsidy and the default assignment of low income enrollees into plans. Removing this distortion could reduce the cost of the program without worsening consumers' welfare. Using data from the the first five years of the program, an econometric model is used to estimate consumers demand for plans and to compute what premiums would be without the subsidy distortion. Preliminary estimates suggest that the reduction in premiums of affected plans would be substantial.