Blog

Health Economics in the Highlands: Key Takeaways from the 5th Annual Highland Health Economics Symposium

July 6, 2026

The 5th Annual Highland Health Economics Symposium (HHES), held June 22–25 in Oban, Scotland, gathered leading health economists to address some of the most pressing questions in modern health policy. Framed around disruptive market innovations, this year’s symposium featured seven research papers exploring how physicians, insurers, manufacturers, and intermediaries shape the cost, quality, and accessibility of care.

Together, the research highlights a crucial reality: health system efficiency relies heavily on aligning incentives across fragmented markets. Whether analyzing billing frictions in private insurance, the long-term savings of curative drug models, or the strategic ripple effects of clinical trial disclosures, HHES provided critical insights for improving value and access.

Below are the core findings and policy takeaways from each of the seven papers. For more information about The HEAL Network’s annual symposiums and future opportunities to participate, please visit www.thehealnetwork.org/annual-conferences/ or contact info@thehealnetwork.org.

Paper 1: Cost and Mortality Effects of Primary Care Physicians

Jason Abaluck, Yale University School of Management and NBER
Zachary Bleemer, Princeton University and NBER
Peter Hull, University of Chicago and NBER
Amanda Starc, Northwestern University, Kellogg School of Management and NBER

Research Question

How much do individual primary care physicians affect patient mortality and health care costs, and can physician quality be measured in a way that is useful for insurance design, regulation, and patient assignment?

Main Findings

The authors develop and validate measures of the health and cost impacts of primary care physicians serving Medicare fee-for-service patients.
Using physician retirements and exits as quasi-experimental variation, the paper shows that risk-adjusted differences in patient mortality capture meaningful causal effects of physician quality (forecast coefficient of 0.71).
A primary care physician who is one standard deviation above the mean in the authors’ quality measure reduces patient mortality by about 13%.
The paper finds that physician characteristics can predict better outcomes at lower cost. In particular, medical school quality and other training characteristics are associated with physicians who deliver better results.
Primary care physicians appear to account for more variation in Medicare Advantage plan impacts than hospitals do, suggesting that PCP quality is a central driver of plan performance.

Key Takeaways

Primary care quality is not interchangeable. Which physician a patient sees can meaningfully affect mortality and spending.
Better measurement of physician performance could help insurers, regulators, and health systems identify high-value primary care.
Physician assignment and network design may be powerful tools for improving outcomes without raising costs.
Traditional health policy debates often focus on hospital quality or insurance plan design, but this paper suggests primary care physicians may deserve more attention as drivers of patient outcomes.

Bottom Line: Improving the match between patients and high-quality primary care physicians could produce meaningful health gains without increasing overall spending. The paper makes a strong case that physician-level quality measurement should play a larger role in health policy.

Paper 2: Fragmented Insurance and Billing Frictions: Understanding Denied Health Insurance Claims

Riley League, Gies College of Business, University of Illinois Urbana-Champaign and NBER
Mark Shepard, Harvard University and NBER
Myles Wagner, The Ohio State University

Research Question

Why are health insurance claims denied so frequently, even when the services are routine or low cost, and what role does insurance fragmentation play in creating billing frictions?

Main Findings

- The paper studies claim denials using data from the Massachusetts All-Payer Claims Database (MA-APCD), covering claims from 2013–2020.
- Denials are common even among private insurers and often involve low-cost, routine services (e.g., blood draws, basic lab panels, standard office visit codes).
- The authors find a negative relationship between the cost of a procedure and its denial rate. Lower-cost claims are more likely to be denied than expensive ones.
- The paper finds no evidence that denials are targeted at higher-cost enrollees: the share of a patient’s claims that are denied is no higher for high-spending patients than low-spending ones, undercutting the idea that denials are used to selectively push out unprofitable patients.
- A substantial share of denials are effectively predictable ex ante: about 3% of all claims are submitted to insurers that deny over 80% of claims for that specific procedure, and these predictable cases account for roughly 28% of all denials.
- The paper shows that claim denials are difficult to explain in a simple model with one provider and one insurer (such a model predicts low, price-increasing denial rates and no billing for non-covered services). Instead, fragmentation across insurers — combined with providers’ limited ability to tailor billing strategies to each individual payer’s rules — appears central to generating the patterns in the data.
- Denial rates spike when patients change insurers (roughly a 5 percentage-point, or ~35%, increase), suggesting that switching coverage disrupts billing routines; the effect fades over time but persists for months.
- Different insurers have very different denial rates for the same service, producing a bimodal pattern (some insurers almost always deny a given procedure, others almost never do). These “bimodal” procedures are common and account for a disproportionate share of all denials.
- Providers with more billing experience have lower denial rates, and vertically integrated provider-insurer pairs also experience fewer denials.

Key Takeaways

Claim denials are not just one-off administrative mistakes; they reflect deeper frictions in a fragmented insurance system rather than deliberate cost-containment or patient-targeting strategies.
The complexity of billing across many insurers creates waste for providers, payers, and patients.
Even low-dollar, routine claims can generate high administrative costs when they must be denied, corrected, appealed, or resubmitted.
Policy reforms that standardize billing rules, harmonize coverage policies, or reduce unnecessary plan switching could reduce waste without reducing access to care.
Provider experience and vertical integration can reduce denials, but relying on experience alone is unlikely to solve the broader fragmentation problem.

Bottom Line: Denied claims are a visible symptom of administrative fragmentation in U.S. health care — not a targeted tool for controlling costs or screening out expensive patients. Reducing billing complexity, particularly around insurer-specific coverage rules and enrollment churn, could be an important path toward lowering waste and improving the patient and provider experience.

Paper 3: Spending to Save? The Subscription Model for Eradicating Hepatitis C in Louisiana

Kevin Callison, Tulane University
Rena M. Conti, Boston University
Jonathan Gruber, Massachusetts Institute of Technology
Jacob Wallace, Yale University

Research Question

Can a state Medicaid program save money and improve access by paying upfront for broad access to high-cost curative drugs?

Main Findings

The paper evaluates Louisiana’s 2019 subscription model for hepatitis C treatment.
Under the model, Louisiana paid a fixed amount for access to direct-acting antivirals (DAAs), which can cure hepatitis C through a short treatment course.
The policy was designed to address a major affordability problem: hepatitis C drugs were highly effective but expensive (averaging $84,000–$94,000 per course), leading most state Medicaid programs to restrict treatment access.
Using Louisiana Medicaid data from 2017–2023, the authors compare spending trends for Medicaid beneficiaries with hepatitis C to comparison groups with diabetes or HIV to control for broader trends.
Expanded access to hepatitis C treatment significantly reduced medical spending among patients with hepatitis C.
Among treated patients, the authors’ Wald estimate implies annual savings of about $5,303.94 per patient, or roughly one-third of baseline HCV-related spending.
Accounting for both direct treatment-related savings and reduced hepatitis C transmission, the authors estimate the program generated $230–250 million in gross savings over its five-year term, with roughly one-quarter of that attributable to reduced transmission.
The program was money-losing in year 1, underscoring that the fiscal benefits of the subscription model emerge over a longer budgeting horizon rather than immediately.
The authors also find that Louisiana’s approach generated greater fiscal benefits than a counterfactual strategy of simply waiting for national DAA prices to fall through market competition.

Key Takeaways

High-cost drugs can sometimes reduce spending when they cure disease and prevent future complications.
Subscription models may help public payers overcome the short-term budget pressure created by expensive but high-value treatments.
The Louisiana case suggests that innovative payment models can expand access while improving fiscal sustainability — though the fiscal payoff may take more than a year to materialize.
Policymakers should evaluate curative therapies over a longer time horizon, especially when treatment prevents hospitalizations, disease progression, or transmission.
The paper provides an important counterexample to the idea that high drug spending is always wasteful; in some cases, the greater waste lies in failing to treat patients early.

Bottom Line: Louisiana’s hepatitis C subscription model shows how a public payer can “spend to save.” Though the program lost money in its first year, it generated an estimated $230–250 million in gross savings over five years — outperforming a strategy of simply waiting for drug prices to fall — suggesting that when a drug is curative and downstream costs are high, paying more upfront can reduce long-term spending and improve patient access.

Paper 4: Disclosure and the Pace of Drug Development

Colleen Cunningham, University of Utah, Eccles School of Business
Florian Ederer, Boston University Questrom School of Business, CEPR, ECGI and NBER
Charles Hodgson, Yale Department of Economics and NBER
Zhichun Wang, Yale Department of Economics

Research Question

Do mandatory clinical trial disclosure rules accelerate innovation by spreading useful information, or can they slow innovation by changing firms’ incentives to invest?

Main Findings

The paper studies the innovation effects of the 2017 FDA Final Rule, which strengthened requirements for pharmaceutical firms to report clinical trial results.
The policy increased disclosure of completed clinical trial results, especially for unsuccessful trials. Before the rule, 68% of completed Phase 2 trials were disclosed within 24 months; after the rule, that figure rose to 78%.
Disclosure of unsuccessful trial results rose from 60% to 75%, while disclosure of successful results rose from 85% to 92%.
The policy also affected firms’ investment timing. Phase 2 clinical trials took longer to complete after the rule, with the median duration rising from about 700 days to around 900 days, and the share completing within two years fell from 20–25% to 15%.
Firms also became slower to move from Phase 1 to later trials. Before the rule, nearly 70% of Phase 1 trials were followed by another trial within a year; after the rule, that share fell below 50%.
At a more granular level, active oncology Phase 2 trials also saw a reduction in the number of active clinical trial sites: trials that had been running for two years had roughly 8 active sites pre-policy versus fewer than 4 post-policy, a decline of more than 50%.
The authors’ most direct evidence for strategic waiting: trial completions and active site counts decline specifically when more related-drug trials are completed but awaiting disclosure, and this sensitivity is significantly stronger after 2017 — consistent with firms delaying investment because they expect more information to arrive.
Separately, the authors show firms do learn from competitors’ results: a one-standard-deviation increase in successful trials of drugs sharing the same mechanism of action raises the likelihood of a firm’s own follow-on trial by 1.65 percentage points, while unsuccessful trials lower it by 1.17 points. The effect reverses for direct competitors in the same therapeutic class — a rival’s success discourages further investment (likely reflecting competitive preemption), while a rival’s failure encourages it.
The authors rule out some alternative explanations, including that the slowdown reflects date manipulation/backdating or a concurrent 2017 international clinical trial guideline change unrelated to the FDA rule.

Key Takeaways

Transparency can improve the information environment, but it can also change firms’ incentives.
Mandatory disclosure may reduce wasteful duplication by helping firms learn from failed trials.
At the same time, if firms expect competitors’ results to become public, they may delay their own investments and wait for more information — and the evidence suggests this is a deliberate, strategic response rather than a mechanical side effect of the policy.
Clinical trial disclosure policy should balance the value of transparency against the risk of slowing the pace of drug development.
The paper highlights a broader policy lesson: disclosure rules can have both ex-post benefits (better-informed R&D decisions) and ex-ante incentive effects (encouraging firms to free-ride and wait).

Bottom Line: Clinical trial transparency is valuable, but not costless. Disclosure rules can improve public knowledge while also encouraging firms to slow investment and free-ride on competitors’ results when the value of waiting outweighs the cost of delay.

Paper 5: Elderly Health and Longevity in the US: Evidence and Implications

Liran Einav, Stanford University and NBER
Amy Finkelstein, Massachusetts Institute of Technology and NBER

Research Question

How have rising life expectancy and improving health among older Americans affected expected lifetime public spending on Medicare and Social Security?

Main Findings

The paper uses nearly three decades of Medicare Current Beneficiary Survey (MCBS) data, from 1992 to 2019, comparing the early period (pooled 1992-1994, referred to as “1993”) to the late period (pooled 2016-2018, referred to as “2017”).
Remaining life expectancy at age 66 increased by about 2.5 years (from 17.5 to 20.0 years).
The entire increase in life expectancy came from additional years in the healthiest morbidity group (zero ADL/IADL limitations) — years in that group rose by 2.6 years, slightly more than the overall increase. This was partly offset by a 0.6-year decline in expected years spent in the worst (most limited) morbidity group. In short, people are not just living longer — the composition of those extra years is healthier, and time spent in serious disability is actually shrinking.
As a result of these demographic changes, expected lifetime Social Security spending rose much faster than Medicare spending under status-quo program rules:
- Social Security: +14% (≈ +$45,200 per 66-year-old)
- Medicare: +6% (≈ +$12,400 per 66-year-old)
- Combined public spending: +≈$57,600, with roughly three-quarters of that increase flowing through Social Security rather than Medicare.
A notable and somewhat counterintuitive finding: expected spending on nursing home and home health care declined slightly over the period, despite people living longer. This is because nursing-home/home-care use is driven mainly by morbidity level (especially the worst morbidity group) rather than age per se, and time spent in that worst group fell.
Increases in life expectancy and lifetime health-care spending were notably larger for men than women, and for higher-income than lower-income individuals.
The paper also builds a stylized life-cycle consumption model to study the optimal allocation of a fixed public budget between Medicare and Social Security. Absent moral hazard, the model implies it’s optimal to direct incremental public dollars toward Medicare first (since it insures against both mortality risk and health risk, and can act as an indirect way to back-load payments as health deteriorates with age). Once moral hazard (excess/inefficient medical spending induced by insurance) is incorporated, the model produces a more realistic interior solution — some balance between Medicare and Social Security — and shows how the optimal balance shifts as demographics change.

Key Takeaways

Longer lives do not automatically translate into proportionally higher Medicare spending — the composition of added years (healthy vs. unhealthy) matters as much as their number.
Because the added years have been disproportionately healthy, population aging over this period has put more fiscal pressure on Social Security than on Medicare.
Nursing home and long-term care costs, contrary to conventional wisdom, have not risen with increased longevity — they track morbidity, not age.
Policy debates about aging should distinguish between longevity (which mainly drives Social Security costs) and morbidity (which mainly drives health-care costs), rather than treating “aging” as a single undifferentiated cost driver.
The paper challenges simple narratives that population aging necessarily produces explosive health-care cost growth.
Public spending debates should weigh not just the total level of spending on the elderly, but the allocation between health insurance (Medicare) and income/longevity support (Social Security) — and the paper’s model suggests that absent other frictions, incremental dollars are generally worth more as insurance (Medicare) than as annuity income (Social Security).

Bottom Line: Americans are living longer, but the additional years have been healthier than many fiscal debates assume. That shifts more of the “cost of aging” onto Social Security than Medicare, and means nursing-home/long-term-care spending isn’t rising in lockstep with longevity the way conventional wisdom suggests.

Paper 6: Price Controls with Imperfect Competition and Choice Frictions: Evidence from Indian Pharmaceuticals

Harsh Gupta, London School of Economics
Shengmao Cao, Kellogg School of Management, Northwestern University

Research Question

How do pharmaceutical price controls affect consumer welfare and overall social welfare when markets are shaped by both imperfect competition and choice frictions?

Main Findings

The paper examines a large-scale pharmaceutical price-control policy in India (the Drug Price Control Order of 2013).
The policy reduced prices for regulated drugs by about 24% and increased sales of those products by about 36%.
The authors find little evidence that the price controls led to significant product entry or exit overall — though exits did occur in about 18% of the therapeutic markets studied, with meaningful welfare consequences in at least one (diabetes).
In a standard welfare analysis, the policy appears to generate meaningful gains in consumer and social surplus in most markets, because price caps reduce the distortions caused by market power.
However, the paper shows that this standard analysis misses an important second distortion: choice frictions.
Using a survey of physicians’ own drug choices, the authors find that consumers systematically overvalue expensive, often multinational, brands — precisely the products that received the largest price cuts under the policy.
Once these choice frictions are incorporated, estimated consumer surplus gains fall by roughly 30% on average (ranging from 6% to 65% smaller, depending on the therapeutic market).
Critically, this isn’t just a “may” — the paper’s structural results show true social welfare declining in all five therapeutic markets studied once choice frictions are accounted for, even though revealed-preference welfare had risen in four of the five.
The authors also evaluate alternative price and non-price regulations (e.g., more granular/brand-level price ceilings, reducing choice frictions directly, government entry, generic substitution, quality standards) that could better address both sources of market failure jointly.

Key Takeaways

Lower drug prices and higher sales do not automatically mean that a policy improves welfare.
Price controls can help when market power leads firms to charge prices above competitive levels.
But when patients or prescribers overvalue certain products, price controls can steer more consumption toward those overvalued products — and the paper’s estimates show this effect was large enough to turn welfare gains into welfare losses in the markets studied.
Pharmaceutical regulation should account for both pricing distortions and information or choice frictions.
Policymakers may need to pair price regulation with non-price reforms, such as better information, prescribing guidance, or quality-focused regulation.
The paper cautions against evaluating drug pricing policy solely by price declines or utilization increases.

Bottom Line: While price controls are intended to mitigate market power, in this setting they failed to improve overall societal welfare because they amplified existing market distortions and consumer choice frictions. The Indian pharmaceutical market demonstrates why drug pricing policies must look beyond short-term patient savings, showing that top-down price caps can cause net welfare losses by inadvertently steering patients away from optimal product choices.

Paper 7: Going for Broker? Intermediation in Health Insurance Markets

Anran Li, Department of Economics, University of Minnesota Twin Cities
Tong Liu, MIT Sloan School of Management
Anthony LoSasso, La Follette School of Public Affairs, University of Wisconsin-Madison
Nicholas Tilipman, Carey Business School, Johns Hopkins University

Research Question

How do insurance brokers affect employer health plan choices, premiums, insurer competition, and overall welfare in the employer-sponsored insurance market?

Main Findings

The paper studies brokers as intermediaries in employer-sponsored health insurance markets, using a novel database of contracting relationships among employers, brokers, and insurers in New York State.
Two market distortions are documented: (1) brokers exhibit agency frictions, steering employers toward plans that pay brokers higher commissions; (2) commission levels shape insurer-broker contracting networks, which in turn affect the degree of competition among insurers.
The authors develop and estimate a structural model of employer demand, insurer pricing, and broker-insurer network formation.
The abstract reports that a one-percentage-point commission cap raises employer surplus ~3% via reduced steering but lowers it >6% via reduced competition, for a net decline of ~3%.

Key Takeaways

Brokers are not neutral pass-through agents; their incentives shape which plans employers choose.
Commission-based compensation can distort plan selection, but the welfare effect of capping commissions is genuinely ambiguous in this model, since it also narrows broker-insurer networks and can soften competition.
Policymakers should weigh not just conflicts of interest but also market-wide competitive effects when designing broker regulation.

Bottom Line: Insurance brokers play a powerful role in employer-sponsored health insurance markets. Reforms that improve broker accountability may help employers, but poorly designed commission limits could reduce competition.