Did the Affordable Care Act’s Closure of the “Donut Hole” Shift Costs or Simply Raise Prices?

Hofmann and Huang (2024) ask whether changes to Medicare Part D—specifically closing the coverage gap—affect drug prices and consumer behavior. They analyze detailed claims data from a 20% sample of Medicare beneficiaries from 2006–2018, covering about 1.9 billion prescription drug purchases. They find that closing the gap increased drug use, with beneficiaries 5 percentage points less likely to skip prescriptions. However, manufacturers responded by raising prices: highly exposed drugs saw price increases of about 21% by 2018. As a result, the intended $100 benefit to consumers fell to about $55, with much of the cost shifted to the government.

Why This Article Was Selected for The Policy Scientist

This article addresses a foundational policy issue: how public insurance expansions interact with private market pricing. That question is especially timely as policymakers revisit drug pricing through recent federal reforms and as other countries confront similar tensions in mixed public–private systems. Hofmann and Huang, who have an established body of work in this area, extend prior research on Medicare Part D by showing that insurance design can shift costs in unintended ways when firms respond strategically. The use of comprehensive national claims data strengthens the analysis, and the quasi-experimental design is credible.

Full Citation and Link to Article

Hofmann, K., & Huang, Z. (2024). The effect of public insurance design on pharmaceutical prices: Evidence from Medicare Part D. American Economic Association Papers & Proceedings / AEA Journal (forthcoming or published version depending on issue). https://www.aeaweb.org/articles?id=10.1257/pol.20250265 

Central Research Question

This article interrogates a central tension in modern health policy: whether expanding public insurance within a privately priced market improves access or instead induces offsetting firm behavior. The specific question is whether the Affordable Care Act’s closure of the Medicare Part D coverage gap altered not only beneficiary drug utilization but also the pricing strategies of pharmaceutical manufacturers. The authors frame this as an equilibrium problem—policy changes that reduce consumer cost exposure may simultaneously weaken price discipline, enabling firms to recapture the intended transfer through higher prices. The core inquiry is therefore not simply whether coverage expansion increases utilization, but whether it reshapes the distribution of costs across beneficiaries, firms, and the federal government.

Previous Literature

The paper builds directly on a well-established literature analyzing nonlinear cost-sharing in Medicare Part D, particularly the canonical finding that beneficiaries respond sharply to the coverage gap by reducing consumption. Prior work has documented bunching at the kink, declines in adherence, and associated health consequences. That literature is primarily demand-focused. A smaller set of studies has examined insurer behavior, formularies, and subsidy design, but relatively little work has provided direct empirical evidence on manufacturer pricing responses to policy-induced changes in cost-sharing.

Hofmann and Huang extend this literature by shifting the analytical lens to the supply side while preserving the demand-side insights that motivate the policy intervention. Their contribution is to connect these strands: the same policy that reduces cost-sharing and increases utilization may simultaneously alter the elasticity of demand faced by firms. This study therefore advances the literature by treating Medicare Part D as a strategic environment, not merely a setting for reduced-form demand estimation. It builds implicitly on prior work on equilibrium effects in insurance markets but moves beyond simulation to direct empirical measurement of price responses.

Data

The empirical analysis is anchored in an exceptionally rich administrative dataset: a 20 percent random sample of Medicare beneficiaries spanning 2006 to 2018, comprising approximately 8.2 million individuals and nearly 1.9 billion prescription drug claims. The data provide granular information on retail prices, out-of-pocket spending, insurer payments, and subsidy flows across coverage phases. This level of detail allows the authors to observe both behavioral responses and price dynamics with precision.

The dataset is further augmented with drug-level characteristics, including patent status, therapeutic class, and brand versus generic classification. This enables the authors to align pricing outcomes with underlying market structure, particularly the presence or absence of competition. The principal limitation—common to this literature—is the absence of rebate data, restricting analysis to retail prices. However, given that beneficiary cost-sharing is tied to retail prices, the data remain highly relevant for assessing distributional effects. Overall, the scale, scope, and administrative origin of the data materially strengthen internal validity and reduce concerns about measurement error.

Methods

The authors employ a two-part empirical strategy. First, they extend standard bunching analysis to quantify how closing the coverage gap altered beneficiary behavior. This provides a direct test of whether the policy achieved its intended demand-side effect—reducing cost-related non-adherence.

Second, and more consequentially, they implement a quasi-experimental event study design to estimate manufacturer pricing responses. The identification strategy exploits cross-drug variation in exposure to the policy, measured by the share of pre-policy revenue derived from non–low-income-subsidy beneficiaries in the coverage gap. Drugs with higher exposure are more affected by both increased demand and the mandated manufacturer discount. The authors then estimate differential price trends for high- versus low-exposure drugs before and after the policy change, controlling for drug-market fixed effects, market-year shocks, and patent lifecycle dynamics.

This approach constitutes a difference-in-differences design with continuous treatment intensity embedded in an event study framework. The reliance on pre-trend validation and rich fixed effects enhances credibility, though the design remains observational. It does not achieve the inferential strength of a randomized controlled trial, nor does it fully eliminate concerns about unobserved heterogeneity correlated with exposure. Nevertheless, within the institutional constraints of the setting, the strategy is well-executed and yields economically meaningful estimates.

Findings/Size Effects

The findings are both internally consistent and substantively large. On the demand side, closing the coverage gap reduced cost-related non-adherence: beneficiaries became approximately 5 percentage points less likely to forgo prescriptions when approaching the gap. The classic bunching response at the coverage threshold declines sharply, with excess mass falling from 2.7 percentage points in 2010 to 1.6 percentage points in 2011 and continuing to diminish thereafter. These results confirm that the policy meaningfully increased drug utilization.

The supply-side response, however, is decisive. By 2018, retail prices for high-exposure drugs are approximately 21 percent higher than for low-exposure drugs, with a monotonic increase over time. This effect is concentrated among brand-name drugs lacking generic competition, consistent with standard models of market power. Notably, the magnitude of the price response exceeds what would be predicted by demand shifts alone, indicating that manufacturers are also responding to the implicit tax imposed by the discount requirement.

The distributional implications are stark. The policy was designed to generate an approximately $100 transfer to beneficiaries, financed by manufacturers. Once endogenous price responses are incorporated, the realized transfer falls to roughly $55. In some cases, beneficiaries who do not enter the coverage gap experience net losses due to higher prices. Moreover, the incidence of the policy shifts toward the federal government, as higher prices increase spending in subsidized segments of the program. The central empirical result is therefore not merely that prices rise, but that the intended redistribution is substantially eroded.

Conclusion

This study demonstrates, with unusual clarity, that public insurance design cannot be evaluated in isolation from firm behavior. The central policy mechanism—reducing cost-sharing to expand access—operates directly on demand, but indirectly on supply by weakening price sensitivity and altering incentives for manufacturers. The result is a partial unraveling of the policy’s intent: increased utilization is accompanied by strategic price increases that reallocate the gains away from beneficiaries.

The broader implication is structural. Any policy that expands insurance coverage within a privately priced market must anticipate equilibrium responses from firms with pricing power. The Medicare Part D reform provides a clean empirical illustration of this principle. The policy succeeded in reducing barriers to access, but it did not fully deliver its intended financial benefits because it underestimated the capacity of firms to adjust prices in response to altered demand conditions.

The strength of the paper lies in its integration of high-quality administrative data with a credible quasi-experimental design to capture these dynamics. While the absence of experimental variation limits definitive causal claims, the magnitude, consistency, and timing of the results provide compelling evidence of strategic pricing behavior. Future research employing stronger causal identification—where feasible—would further solidify these conclusions, but the core insight is unlikely to be overturned.

At a broader level, the article reframes how policy analysts should evaluate insurance expansions. The relevant question is not only whether coverage increases utilization, but whether the institutional design preserves or dissipates the intended transfer once market responses are accounted for. In this respect, the paper makes a substantive contribution: it shifts the analytical focus from partial equilibrium gains to general equilibrium incidence. That shift is essential for designing durable policy in sectors characterized by concentrated market power and complex regulatory structures.