What the Medicare Part D benefit structure does to medication use

The standard Medicare Part D benefit does not present beneficiaries with a single price for their medications. Beneficiaries instead move through a sequence of coverage phases over the course of the year. They pay a deductible, then a share of drug costs during the initial coverage phase, then a considerably larger share in what was known as the coverage gap or donut hole, and finally a small share once catastrophic coverage begins. Out-of-pocket costs therefore change discontinuously at each phase boundary, and the entire cycle resets on January 1 of the following year.

This structure raises a question that turns out to be more interesting than it first appears. Because the phase boundaries are in principle foreseeable, standard economic theory predicts that beneficiaries should smooth their medication use across the year, taking into account the price they expect to face later rather than responding only to the price in front of them. Whether people actually behave this way has implications well beyond Part D, since many insurance benefits vary cost sharing over the course of a year.

Beneficiaries do anticipate price changes

Working with Yuting Zhang, I approached this using variation in beneficiaries' birth months. Beneficiaries who enroll in Part D when they first turn 65 and who have birthdays earlier in the calendar year have more of the year remaining, and are therefore more likely to reach the coverage gap, than those who enroll later in the year. That variation is plausibly exogenous, and comparing the two groups over their first three months of enrollment allows us to ask whether expectations about a future price change affect consumption in the present.

We found strong evidence of anticipatory behavior, with an implied elasticity with respect to future prices ranging from roughly −0.2 to −0.5. We also found that beneficiaries adjusted primarily by changing the quantity of prescriptions they filled rather than by substituting between brand-name and generic drugs, which suggests that the response operates through the amount of medication obtained rather than through a shift toward cheaper alternatives.

The reset still matters

A related question concerns what happens at the annual reset itself. Beneficiaries who receive low-income subsidies provide a useful comparison group, since many of them face zero or small fixed copayments throughout the year and so never encounter the phase structure. Comparing against that group, we found that beneficiaries delay reinitiating medications in December and are significantly more likely to reinitiate in January than in other months. Where reinitiation declined in the final months of the year, the pattern was driven largely by those facing higher prices as a result of the coverage gap.

What closing the gap changed

The Affordable Care Act gradually eliminated the coverage gap between 2011 and 2020, which provides a useful natural experiment. With Jinan Liu and Yuting Zhang, I estimated the effects on out-of-pocket spending and medication use, and the estimation proved harder than it initially appears. Over the same period a number of blockbuster drugs commonly used by the Medicare population came off patent, and the resulting generic entry independently reduced prices, so that attributing the full change to the policy would overstate its effect. Because generic entry affected different therapeutic classes at different times, we ran difference-in-differences models by therapeutic category at the beneficiary-month level in order to separate the two.

We found that filling the gap substantially reduced out-of-pocket spending and increased the use of branded drugs, which carried larger discount rates during the period we studied. Consistent with the design of the policy, the reductions in out-of-pocket spending were largest among beneficiaries who actually reached the gap, who were at older ages, or who had coexisting chronic conditions. The methodological result is worth stating on its own, since it applies well beyond this particular policy. Without accounting for generic entry, the effect of filling the coverage gap is underestimated for branded drugs and overestimated for generic drugs, so that any evaluation ignoring patent expiration will get both directions wrong.

Structure as well as generosity

Discussions of prescription drug coverage tend to focus on generosity, meaning how much of the cost the plan bears. The evidence from Part D suggests that structure matters a great deal as well, since a benefit delivering the same average subsidy through a smooth price would not produce anticipatory adjustment ahead of the gap or a surge in reinitiation each January. The Part D benefit continues to evolve, including the out-of-pocket cap established under the Inflation Reduction Act. The relevant questions concern not only how much beneficiaries are protected, but also at what points in the year their prices change and how well they are able to anticipate those changes.