Preferred Providers
The Rise of Healthcare Jobs
Joshua Gottlieb et al.
NBER Working Paper, March 2025
Abstract:
Healthcare employment has grown more than twice as fast as the labor force since 1980, overtaking retail trade to become the largest industry by employment in 2009. We document key facts about the rise of healthcare jobs. Earnings for healthcare workers have risen nearly twice as fast as those in other industries, with relatively large increases in the middle and upper-middle parts of the earnings distribution. Healthcare workers have remained predominantly female, with increases in the share of female doctors offsetting increases in the shares of male nurses and aides. Despite a few high-profile examples to the contrary, regions experiencing manufacturing job losses have not systematically reinvented themselves by pivoting from "manufacturing to meds."
Policy options for the drug pricing conundrum
Kate Ho & Ariel Pakes
Proceedings of the National Academy of Sciences, 4 March 2025
Abstract:
Current proposals aimed at reducing U.S. pharmaceutical prices would have immediate benefits (particularly for low-income and elderly populations), but could dramatically reduce firms’ investment in potentially highly welfare-improving Research and Development (R&D). The United States subsidizes the worldwide pharmaceutical market: U.S. drug prices are more than 250% of those in other Organization for Economic Co-operation and Development (OECD) countries. If each drug had a single international price across the highest-income OECD countries and total pharmaceutical firm profits were held fixed: U.S. prices would fall by half; every other country’s prices would increase (by 28 to over 300%); and R&D incentives would be maintained. We propose a potential lever for the U.S. government to influence worldwide drug pricing: access to the Medicare market.
The Economics of Healthcare Fraud
Jetson Leder-Luis & Anup Malani
NBER Working Paper, March 2025
Abstract:
Healthcare fraud imposes a sizable cost on U.S. public healthcare budgets and distorts health care provision. We examine the economics of health care fraud and enforcement using theory and data and connect to a growing literature on the topic. We first offer a new economic definition of health care fraud that captures and connects the wide range of activities prosecuted as fraud. We define fraud as any divergence between the care an insurer says a patient qualifies for, the care a provider provides, and the care a provider bills for. Our definition clarifies the economic consequences of different categories of fraud and provides a framework for understanding the slate of existing studies. Next, we examine the incentives for committing and for prosecuting fraud. We show how fraud is driven by a combination of inadequate (expected) penalties for fraud and imperfect reimbursement rates. Public anti-fraud litigation is driven by the relative monetary, political or career returns to prosecuting fraud and by prosecutorial budgets. Finally, we examine the prevalence of health care fraud prosecutions across types of fraud and types of care, and across the US, by machine learning on text data from Department of Justice press releases.
Randomized Trial of a Generative AI Chatbot for Mental Health Treatment
Michael Heinz et al.
NEJM AI, March 2025
Background: Generative artificial intelligence (Gen-AI) chatbots hold promise for building highly personalized, effective mental health treatments at scale, while also addressing user engagement and retention issues common among digital therapeutics. We present a randomized controlled trial (RCT) testing an expert–fine-tuned Gen-AI–powered chatbot, Therabot, for mental health treatment.
Methods: We conducted a national, randomized controlled trial of adults (N=210) with clinically significant symptoms of major depressive disorder (MDD), generalized anxiety disorder (GAD), or at clinically high risk for feeding and eating disorders (CHR-FED). Participants were randomly assigned to a 4-week Therabot intervention (N=106) or waitlist control (WLC; N=104). WLC participants received no app access during the study period but gained access after its conclusion (8 weeks). Participants were stratified into one of three groups based on mental health screening results: those with clinically significant symptoms of MDD, GAD, or CHR-FED. Primary outcomes were symptom changes from baseline to postintervention (4 weeks) and to follow-up (8 weeks). Secondary outcomes included user engagement, acceptability, and therapeutic alliance (i.e., the collaborative patient and therapist relationship). Cumulative-link mixed models examined differential changes. Cohen’s d effect sizes were unbounded and calculated based on the log-odds ratio, representing differential change between groups.
Results: Therabot users showed significantly greater reductions in symptoms of MDD (mean changes: −6.13 [standard deviation {SD}=6.12] vs. −2.63 [6.03] at 4 weeks; −7.93 [5.97] vs. −4.22 [5.94] at 8 weeks; d=0.845–0.903), GAD (mean changes: −2.32 [3.55] vs. −0.13 [4.00] at 4 weeks; −3.18 [3.59] vs. −1.11 [4.00] at 8 weeks; d=0.794–0.840), and CHR-FED (mean changes: −9.83 [14.37] vs. −1.66 [14.29] at 4 weeks; −10.23 [14.70] vs. −3.70 [14.65] at 8 weeks; d=0.627–0.819) relative to controls at postintervention and follow-up. Therabot was well utilized (average use >6 hours), and participants rated the therapeutic alliance as comparable to that of human therapists.
Expert Patients’ Use of Avoidable Health Care
Pragya Kakani, Simone Matecna & Amitabh Chandra
NBER Working Paper, March 2025
Abstract:
We measure whether expert patients -- those trained as physicians and nurses -- have fewer emergency department visits and the reasons for these differences. Relative to similar patients physicians and nurses had 19.8% and 5.1% fewer ED visits, principally due to fewer avoidable visits. The differences in avoidable visits between physicians and other patients were largest for diagnoses commonly requiring prescriptions, which physicians often self-prescribed. Our results suggest that improving access to prescriptions for acute symptoms, more than improving patient education, may reduce avoidable health care.
Artificial Intelligence on Call: The Physician's Decision of Whether to Use AI in Clinical Practice
Tinglong Dai & Shubhranshu Singh
Journal of Marketing Research, forthcoming
Abstract:
Physicians are increasingly able to use artificial intelligence (AI) systems to aid their medical decision-making. This paper examines a physician’s decision regarding whether to use an assistive AI system when prescribing a treatment plan for a patient. Using AI helps the physician generate an informative signal that lessens clinical uncertainty. It can also change the physician’s legal liability in the event of patient harm. We analyze two patient-protection schemes that determine physician liability when using AI: the prevailing patient-protection scheme uses the AI signal to enforce the current standard of care, whereas an emerging scheme proposes using the AI signal as the new standard of care. We show that in both schemes, the physician has an incentive to use AI in low-uncertainty scenarios, even if AI provides little value. Furthermore, the physician may avoid using AI in higher-uncertainty scenarios where AI could have aided in better decision-making. As AI becomes more precise, the physician may become more hesitant to use it on certain patients. A comparison of the physician’s decision to use AI under the two schemes reveals that using the AI signal as the new standard of care may mitigate AI underuse (overuse) for certain patients but may boost AI underuse (overuse) for some other patients.
The Rise and Fall (and Rise) of the Affordable Care Act: Varying Impacts on Coverage Over Time and Place
Gabriella Aboulafia, Jonathan Gruber & Benjamin Sommers
NBER Working Paper, March 2025
Abstract:
The Affordable Care Act (ACA) significantly expanded health insurance in the United States, but its impact has varied across time and states. We assess the law’s heterogeneous impacts over the three presidential administrations since its enactment, as well as across states with different levels of implementation of the law. We focus on Medicaid expansion and Marketplace subsidies, including the enhanced subsidies under the American Rescue Plan of 2021 (ARP). We use national household survey data and a triple-difference design -- leveraging variation by time, state, and income -- to identify the coverage impacts of the key components of the law. We find that 55% of ACA-related coverage gains between 2013 and 2023 came from Marketplace subsidies -- about 37% from the original ACA subsidies and 19% from the ARP enhancements -- while 45% were due to Medicaid, including from the "welcome mat" effect. Coverage gains differed substantially across presidential administrations, with Marketplace subsidies proving roughly 30% more effective under Presidents Obama and Biden than under President Trump. The same subsidy amount was more than twice as effective in states with their own Marketplaces than in states relying on the federal Marketplace. Our findings highlight that while the ACA’s explicit economic features drive coverage gains, their effectiveness can be substantially enhanced or hindered through federal and state implementation.
Subscriptions to Prescriptions: Lessons from Louisiana’s Effort to Eliminate Hepatitis C
James Flynn, Bethany Lemont & Barton Willage
NBER Working Paper, March 2025
Abstract:
Hepatitis C is a major public health concern due to its high rates of infection and mortality. Recent breakthroughs in pharmaceuticals not only have the potential to cure hepatitis C but could also cause large positive health externalities through reduced transmission. The high cost of these drugs under traditional reimbursement schemes create large obstacles to care, but a recent first-of-its-kind two-part tariff system in Louisiana aims to circumvent these obstacles using a modified subscription model with an exclusive pharmaceutical provider. Under this model, the medication is provided at no marginal cost to the state to cover the state's Medicaid and incarcerated population. This creates an incentive for Louisiana to aggressively test and treat as many patients as possible in order to maximize the benefits of this agreement. Using a number of different data sources, we implement synthetic control and event-study specifications, and find that detection and treatment of hepatitis C increased dramatically, with meaningful reductions in hepatitis C-related mortality and liver transplants after this agreement. Finally, after calculating the Marginal Value of Public Funds of this agreement, we find that the program more than pays for itself.