Does ACA Medicaid Expansion Fail to Improve Health While Straining State Budgets? What the Oregon Experiment, Mortality Records, and Fiscal Audits Show

More than a decade after the Affordable Care Act's major coverage provisions took effect in January 2014—following the U.S. Supreme Court's landmark 2012 ruling in NFIB v. Sebelius, which made Medicaid expansion optional for states—forty states and the District of Columbia have adopted expansion, while ten states (led by Texas, Florida, Georgia, Wisconsin, Tennessee, Mississippi, Alabama, South Carolina, Kansas, and Wyoming) continue to decline it ^{[6], [9]}. In Washington, D.C., and across state capitals in the South and Midwest, the debate over Medicaid expansion remains one of the most consequential flashpoints in American domestic policy. Throughout 2025 and 2026, congressional Republicans and Trump administration health advisors—guided by policy blueprints from the Paragon Health Institute (founded by former White House economic advisor Brian Blase), the Foundation for Government Accountability (FGA), and the Heritage Foundation—have targeted the ACA's Medicaid expansion funding structure as a centerpiece of federal budget reconciliation and entitlement reform ^{[7], [8]}.

At the heart of the conservative case against Medicaid expansion lies a provocative twofold argument. First, conservative policymakers contend that expanding a public safety-net program designed in 1965 for low-income children, pregnant women, frail seniors, and individuals with severe disabilities to cover more than 21 million working-age, able-bodied adults (earning up to 138% of the Federal Poverty Level) has bloated federal outlays, generated tens of billions in improper payments, and created a perverse federal matching subsidy that favors able-bodied adults over the disabled ^{[7], [8]}. Second, critics argue that Medicaid coverage itself delivers remarkably poor clinical value—repeatedly invoking the Oregon Health Insurance Experiment (OHIE) to assert that Medicaid expansion neither improves physical health outcomes nor keeps low-income patients out of hospital emergency rooms ^{[1], [2], [8]}.

"When Oregon ran the only randomized controlled trial of Medicaid in modern American history, researchers found that putting uninsured adults on Medicaid did not significantly improve blood pressure, cholesterol, or diabetic blood sugar, while increasing emergency room visits by 40 percent. Yet because Washington pays 90 cents on the dollar for able-bodied expansion adults—while paying as little as 50 cents for a child with cerebral palsy or an elderly nursing-home resident—states have been bribed into expanding a program rife with improper payments." — Synthesis of Conservative Health Policy Critiques (Paragon Health Institute, Foundation for Government Accountability, and Non-Expansion State Legislative Audits) ^{[1], [2], [7], [8]}

Do these claims hold up under rigorous econometric and clinical scrutiny? To separate partisan rhetoric from evidence, The Dispatch audited the complete series of peer-reviewed papers from the Oregon Health Insurance Experiment, subsequent population-wide mortality studies in the Quarterly Journal of Economics, hospital financial datasets from Health Affairs and the UNC Sheps Center, and federal fiscal accounting from the Congressional Budget Office (CBO), the Medicaid and CHIP Payment and Access Commission (MACPAC), and the Centers for Medicare & Medicaid Services (CMS) ^{[1], [3], [5], [6], [7]}.

Where the Conservative Critique Has Merit: Emergency Room Utilization, Enrollment Overshoots, and the 90% FMAP Distortion

A balanced, evidence-based examination must begin by acknowledging that conservative health economists and fiscal analysts are right on three substantial empirical points that proponents of the Affordable Care Act initially downplayed or mispredicted.

1. The Emergency Room Paradox: Why Insurance Increased ED Visits by 40%

When Congress debated the Affordable Care Act in 2009 and 2010, bipartisan supporters and the Obama administration frequently promised that expanding Medicaid would reduce overcrowded hospital emergency rooms by giving uninsured Americans access to primary-care doctors. In January 2014, however, economists Sarah Taubman, Heidi Allen, Bill Wright, Katherine Baicker, and Amy Finkelstein published a landmark study in Science analyzing hospital administrative records from the 2008 Oregon Health Insurance Experiment—in which Oregon used a random lottery among ~90,000 low-income uninsured adults on a waiting list to fill 10,000 open slots in the Oregon Health Plan (Medicaid) ^{[1], [2]}.

Comparing lottery winners against the randomized control group in the Portland metropolitan area over an 18-month period, the researchers found the exact opposite of the political promise: gaining Medicaid coverage increased emergency department utilization by 0.41 visits per person, or 40% relative to the control group average of 1.02 visits (p < 0.001) ^[2]. Furthermore, the surge in ER visits persisted across almost all clinical categories, including off-hours visits and conditions classified as non-emergent or treatable in a primary-care setting—even as Medicaid simultaneously increased outpatient office visits by 35% to 50% ^{[1], [2]}. In basic microeconomic terms, conservatives were proven right about the law of demand: when the out-of-pocket price of visiting a 24-hour, one-stop diagnostic emergency room drops from thousands of dollars to near zero, patients use more of both primary care and emergency room care rather than substituting one for the other ^[2].

2. The 90% Federal Match Rate and Fiscal Moral Hazard

Second, conservative policy scholars—most notably Brian Blase at the Paragon Health Institute and analysts at the Foundation for Government Accountability (FGA)—correctly identify a major structural distortion built into the ACA's financing architecture ^{[7], [8]}. For traditional Medicaid populations established in 1965 (children in poverty, pregnant women, low-income seniors, and individuals with severe physical or intellectual disabilities), the federal government pays a statutory Federal Medical Assistance Percentage (FMAP) ranging from 50% in wealthier states to ~77% in the poorest states, averaging roughly 57% to 60% nationwide ^{[6], [7]}. For the ACA expansion group (non-disabled, working-age adults up to 138% FPL), however, Congress set a permanent 90% federal match rate (after paying 100% from 2014 to 2016) ^{[6], [7]}.

This disparity creates two documented fiscal distortions:

• Marginal Dollar Skew & Provider-Tax Financing: Because a state governor must spend $0.40 to $0.50 of state tax money to draw down $1.00 of Medicaid care for a disabled child or nursing-home resident, but only $0.10 of state funds (often raised through hospital and managed-care "provider taxes" rather than general revenues) to draw down $1.00 for an expansion adult, states face an overwhelming fiscal incentive to maximize expansion enrollment and managed-care capitation rates at federal taxpayers' expense ^{[6], [7]}.
• Enrollment Overshoots and Eligibility Verification Gaps: Across the first decade of implementation, FGA and Paragon audits of CMS enrollment records demonstrated that actual ACA expansion enrollment routinely exceeded initial state and CBO projections by 50% to over 100%, growing to more than 21 million to 23 million adults and driving total federal-state Medicaid outlays past $880 billion to $900+ billion annually ^{[6], [7], [8]}. Moreover, CMS's own Payment Error Rate Measurement (PERM) audits and post-pandemic "unwinding" data (during which states disenrolled over 24 million people between 2023 and 2024, roughly 70% for procedural/paperwork reasons and millions for income ineligibility) confirmed that automated ex-parte renewals and lax income verification contributed to tens of billions of dollars in federal improper payments ^{[7], [8]}.

What the Data Shows Part I: What the Oregon Health Insurance Experiment Actually Found—and Why It Couldn't Measure Mortality

While conservative critics are right about emergency department demand and the 90% FMAP structure, the frequent claim that the Oregon Health Insurance Experiment "proved Medicaid produces zero health benefits" is a classic case of selective citation and statistical misinterpretation ^{[1], [3]}.

1. Inside the May 2013 New England Journal of Medicine Study

In their famous May 2013 NEJM paper ("The Oregon Experiment — Effects of Medicaid on Clinical Outcomes"), Katherine Baicker, Sarah Taubman, Heidi Allen, Bill Wright, and Amy Finkelstein conducted in-person clinical screenings (measuring blood pressure, total and HDL cholesterol, glycated hemoglobin [HbA1c] for diabetes, and PhQ-8 depression scores) on 12,229 lottery participants in the Portland area roughly two years after the 2008 lottery ^[1]. Here is the complete record of what the randomized trial found:

• Substantial Preventive Care, Diabetes Detection, and Mental Health Gains: Medicaid coverage significantly increased the probability of having a regular office-based place of care (+23.8 percentage points, or +50%), increased preventive screenings (mammograms up 60%, cholesterol checks up 20%, Pap smears up 45%), increased the clinical diagnosis of diabetes by 3.83 percentage points (+70%), and increased the use of diabetes medications by 5.43 percentage points ^[1]. Most strikingly, Medicaid coverage reduced the prevalence of positive screening for clinical depression (PHQ-8 ≥ 10) by 9.15 percentage points—a massive 30% relative reduction (p = 0.003)—while significantly improving self-reported physical and mental health ^[1].
• Near-Total Protection Against Medical Financial Ruin: Medicaid reduced the probability of having any out-of-pocket medical spending by 15.3 percentage points, reduced the probability of having to borrow money or skip paying other bills due to medical expenses by 14.2 percentage points (-58%), and reduced catastrophic out-of-pocket medical expenditures (exceeding 30% of household income) by 4.5 percentage points—an reduction of more than 80% (nearly eliminating catastrophic medical financial risk) ^[1].
• The Two-Year Biometric Confidence Intervals: On the three physical biometric lab tests—hypertension (elevated blood pressure), hypercholesterolemia (high cholesterol), and diabetic HbA1c ≥ 6.5%—the point estimates favored the Medicaid group (e.g., a 1.33 percentage-point reduction in hypertension), but the effects were not statistically significant ^[1]. Why? As the authors explicitly explained in the NEJM text, only a tiny fraction of the 12,229 low-income adults (average age ~40) had baseline uncontrolled diabetes (~5%) or hypertension (~16%), meaning the two-year effective sample size for chronic-disease biometrics was only a few hundred individuals—leaving wide 95% confidence intervals (for blood pressure, the confidence interval spanned from a -7.1 percentage-point reduction to a +4.4 percentage-point increase, which could not statistically rule out the entire clinical benefit typically seen in clinical trials of blood-pressure drugs) ^{[1], [3]}.

2. The Statistical Power Problem: Why 12,000 Young Adults Cannot Measure Death Rates

Even more importantly, in their 2012 Quarterly Journal of Economics paper examining one-year Oregon administrative records (N = 74,922), Finkelstein et al. noted that because annual mortality among working-age adults (mean age ~40) is extremely low (~0.8% per year), a sample of a few tens of thousands of people followed for 12 to 24 months has virtually zero statistical power to detect realistic changes in mortality ^{[1], [3]}. Claiming that the Oregon lottery proved Medicaid does not reduce mortality is akin to flipping a coin ten times, observing six heads, and declaring that a 5% bias is impossible: the sample size was simply orders of magnitude too small to measure death rates ^[3].

What the Data Shows Part II: Million-Person Census and IRS Studies Prove Medicaid Expansion Saves Lives

To overcome the sample-size limits of the Portland lottery, health economists turned to massive federal administrative datasets linking millions of Americans across tax, survey, and Social Security death records. Across multiple quasi-experimental and randomized designs, the empirical consensus is unambiguous: health insurance in general, and ACA Medicaid expansion in particular, significantly reduces adult mortality ^{[3], [4]}.

1. Miller, Johnson, and Wherry (Quarterly Journal of Economics, 2021)

In a landmark 2021 study published in the Quarterly Journal of Economics ("Medicaid and Mortality: New Evidence from Linked Survey and Administrative Data"), economists Sarah Miller (University of Michigan), Norman Johnson (U.S. Census Bureau), and Laura R. Wherry (NYU) linked individual-level responses from the Census Bureau's American Community Survey (ACS) to official Social Security Administration Numident death records covering roughly 3.7 million low-income, near-elderly adults aged 55 to 64 between 2008 and 2017 ^[3].

Using a rigorous difference-in-differences design comparing low-income adults in expansion states against identical cohorts in non-expansion states before and after January 2014, Miller et al. documented that:

• 0.132 Percentage-Point Drop in Population Mortality (-9.4% Treatment-on-the-Treated): Across the entire low-income target population in expansion states, annual mortality fell by 0.132 percentage points by the fourth year of expansion—with zero pre-2014 trend differences between expansion and non-expansion states ^[3]. Scaling by the share of the group that actually gained Medicaid coverage (the local average treatment effect), enrolling in Medicaid reduced an individual's annual probability of death by 9.4% ^[3].
• Driven by Treatable Internal-Medicine Diseases: Crucially, falsification tests showed that the mortality reduction was concentrated entirely in internal, disease-related causes of death (such as cardiovascular disease, diabetes complications, cancer, and respiratory disease amenable to prescription drugs and medical management), with no change in external deaths (such as car crashes or homicides) and no effect among high-income adults or seniors over 65 already on Medicare ^[3].
• 19,200 Lives Saved vs. 15,600 Lives Lost in Non-Expansion States: In aggregate, ACA Medicaid expansion saved roughly 4,800 lives per year (or 19,200 lives across 2014–2017) among adults aged 55–64 alone—while the decision of non-expansion states to opt out cost an estimated 15,600 preventable deaths over the same four-year window ^[3].

2. The Randomized IRS Outreach Experiment (Goldin, Lurie, & McCubbin, QJE, 2021) & Pre-ACA Expansions (Sommers et al., NEJM, 2012)

Could Miller et al.'s difference-in-differences findings be an artifact of state-level trends? In the very same 2021 volume of the Quarterly Journal of Economics, Stanford economist Jacob Goldin and U.S. Treasury economists Ithai Z. Lurie and Janet McCubbin published results from a nationwide randomized controlled trial of 3.9 million uninsured taxpayers conducted by the Internal Revenue Service in 2017 ^[4]. The IRS randomly assigned 85% of taxpayers who had paid the ACA individual-mandate penalty to receive an informational letter about enrolling in Medicaid or ACA marketplace coverage, holding 15% as an untouched control group. Among middle-aged adults (ages 45–64), the randomized letter increased health insurance coverage (driven heavily by Medicaid enrollment) and statistically significantly reduced two-year all-cause mortality—estimating that one life was saved for roughly every 1,548 adult months (or ~129 years) of induced insurance coverage ^[4]. These findings align closely with earlier state-expansion studies by Harvard health economist Benjamin D. Sommers, Katherine Baicker, and Arnold M. Epstein (NEJM, 2012), who found a 6.1% reduction in all-cause adult mortality (19.6 fewer deaths per 100,000 adults) following pre-ACA Medicaid expansions in New York, Maine, and Arizona ^[4], as well as Borgschulte and Vogler (Review of Economics and Statistics, 2020), who documented significant county-level mortality reductions across all ACA expansion states ^[3].

The Full Picture: Rural Hospital Survival, Uncompensated Care, and State General-Fund Budgets

Beyond individual clinical outcomes, the debate over Medicaid expansion hinges on two macro-level economic questions: what does expansion do to the survival of rural hospitals, and does the 10% state share crowd out core state budget priorities like K–12 schools and roads?

1. The Rural Hospital Lifeline: Uncompensated Care Cut in Half

In rural America—where populations are older, incomes are lower, and fewer employers offer commercial health insurance—hospitals operate on razor-thin margins and are legally required under the federal Emergency Medical Treatment and Labor Act (EMTALA) of 1986 to treat every emergency patient regardless of ability to pay. When low-income working adults lack coverage, rural hospitals must absorb their treatment as unpaid charity care and bad debt.

In a landmark 2018 study in Health Affairs ("Understanding The Relationship Between Medicaid Expansions And Hospital Closures"), economists Richard C. Lindrooth, Marcelo Perraillon, Rose Hardy, and Gregory Tung analyzed American Hospital Association and CMS Medicare Cost Report financial data across more than 4,600 U.S. hospitals from 2008 to 2016 ^[5]. Their econometric findings were stark:

• Uncompensated Care Slashed by ~50%: In Medicaid expansion states, hospital uncompensated care costs plummeted from roughly 4.1%–6.0% of total hospital operating expenses to 2.0%–3.1% within two years—saving hospitals over $6 billion annually in bad debt ^{[5], [6]}.
• 62% to 84% Lower Risk of Rural Hospital Closure: Medicaid expansion improved hospital operating margins and reduced the probability of hospital closure by 84% overall (0.33 fewer closures per 100 hospitals per year), with the largest protective effects occurring in rural markets (-62% to -84% closure risk) and counties with high pre-2014 uninsurance rates ^[5].
• 74% of Rural Hospital Closures Occurred in Non-Expansion States: Longitudinal tracking by the Cecil G. Sheps Center for Health Services Research at the University of North Carolina through 2025–2026 confirms that of the more than 150 to 190 rural hospitals that have closed or converted away from inpatient acute care since 2010, roughly 74% were located in non-expansion states—led by Texas (26+ closures), Tennessee (16+), Georgia (10+), Oklahoma, Mississippi, and Alabama ^{[5], [9]}. When a rural hospital closes, peer-reviewed studies show local inpatient mortality for time-sensitive emergencies (such as heart attacks and strokes) rises significantly and county per-capita income drops ^[5].

2. State Budgets: Why the 10% State Share Often Produced Net General-Fund Savings

What about the conservative warning that paying the 10% state share of Medicaid expansion would bankrupt state budgets and crowd out K–12 education and highway construction? Comprehensive fiscal audits by the non-partisan congressional Medicaid and CHIP Payment and Access Commission (MACPAC), the Kaiser Family Foundation (KFF), and economist Bryce Ward (Commonwealth Fund) analyzing National Association of State Budget Officers (NASBO) expenditure data from 2010 through 2025 reveal a surprising fiscal reality ^{[6], [9]}:

• Three Built-In State Budget Offsets: When a state adopts ACA Medicaid expansion (paying 10% of costs while receiving 90% in federal funds), it captures three immediate state general-fund savings that frequently exceed the 10% state share:
  1. Replacing 100% State-Funded Programs: Prior to expansion, states spent billions of 100% state and county general-fund dollars on community behavioral health, substance-abuse treatment, indigent charity care pools, and inpatient state prison/jail hospitalizations (which qualify for 90% federal Medicaid matching once an inmate is admitted to an outside hospital for 24+ hours) ^{[6], [9]}.
  2. Transitioning Traditional Categorical Groups to the 90% Match: Certain populations previously covered under optional state Medicaid waivers at the traditional 50%–70% FMAP (such as medically needy adults or breast/cervical cancer screening enrollees) shifted into the 90% expansion match, saving states 20 to 40 cents on every dollar ^[6].
  3. Induced State Tax & Provider-Fee Revenue: The influx of tens of billions of federal healthcare dollars generates state income, sales, and insurer premium-tax revenues, while over 35 expansion states finance most or all of their remaining 10% share through hospital or managed-care assessments endorsed by state hospital associations ^{[6], [9]}.
• Zero Crowd-Out of K–12 Education or Transportation: Econometric panel studies of NASBO state budget data (Sommers & Gruber, Health Affairs; Ward, 2020; KFF, 2021–2025) confirm that state general-fund spending on Medicaid grew no faster in expansion states than in non-expansion states after 2014, and expansion had zero statistically significant negative effect ("crowd-out") on state general-fund spending for K–12 education, higher education, or transportation infrastructure ^{[6], [9]}. In conservative states that expanded Medicaid—such as Indiana, Montana, Louisiana, Virginia, and North Carolina (which expanded in December 2023)—legislative fiscal bureaus documented hundreds of millions of dollars in net state general-fund savings ^{[6], [9]}.

3. What About the "Disabled Crowd-Out / Waiting List" Claim?

Conservative groups like the Foundation for Government Accountability frequently claim that ACA Medicaid expansion "pushed over 600,000 disabled Americans onto Medicaid waiting lists" ^[8]. However, audits by MACPAC and KFF show that this claim conflates two legally distinct programs: anyone who meets federal SSI disability criteria is automatically entitled to regular acute-care Medicaid with no waiting list in all 50 states ^[6]. The waiting lists cited by FGA are specifically for optional Section 1915(c) Home and Community-Based Services (HCBS) waivers (where states cap slots for in-home long-term care aides). Crucially, KFF and MACPAC data prove that the largest HCBS waiting lists in the nation are actually in non-expansion states—most notably Texas, which alone accounts for over 300,000+ of the ~690,000 national HCBS waitlist slots (~45% of the entire U.S. total), followed by Florida—and that expansion states actually increased per-enrollee spending and slot availability for HCBS disability waivers faster than non-expansion states after 2014 ^{[6], [9]}.

Conclusion: A Real Federal Fiscal Distortion—and Real Lives Saved

When over a decade of randomized clinical data, administrative mortality records, and fiscal audits are evaluated without partisan filters, the debate over ACA Medicaid expansion resolves into a clear, nuanced trade-off between federal fiscal discipline and measurable health and hospital protection.

Conservative health economists at the Paragon Health Institute and legislative hawks in Washington are empirically justified in pointing out that Medicaid expansion was not a free lunch for the federal Treasury ^{[2], [7]}. As the randomized Oregon Health Insurance Experiment proved, eliminating out-of-pocket costs increased emergency department visits by 40% rather than emptying hospital waiting rooms ^[2]. And at the macro-fiscal level, the ACA's statutory formula—reimbursing states at 90% FMAP for working-age, able-bodied adults while reimbursing them at an average of only ~57% for low-income children, pregnant women, and people with severe disabilities—created genuine fiscal moral hazard, fueled provider-tax financing schemes, and contributed to massive federal enrollment overshoots and improper eligibility payments ^{[6], [7], [8]}.

Yet the accompanying conservative claim that Medicaid expansion "fails to improve health, kills rural budgets, and crowds out state schools" does not survive empirical scrutiny ^{[1], [3], [5]}. Even within two years, the Oregon experiment demonstrated a 30% drop in clinical depression, a 70% increase in diabetes diagnosis, and the near-elimination of catastrophic medical debt ^[1]. Over longer horizons and population-wide samples of 3.7 million to 3.9 million Americans, linked Census, Social Security, and randomized IRS studies prove beyond serious doubt that Medicaid expansion reduced annual mortality among low-income near-elderly adults by 9.4%—saving thousands of lives every year from treatable cardiovascular, metabolic, and oncological diseases—while slashing rural hospital closures by more than 60% and generating net savings for state general funds ^{[3], [4], [5], [9]}. For policymakers in 2026, the evidence points not toward denying coverage to low-income uninsured workers in non-expansion states, but toward pairing coverage access with tighter eligibility verification, primary-care co-pay incentives to curb non-emergent ER use, and a more balanced federal matching structure that treats traditional disabled beneficiaries at least as generously as able-bodied adults ^{[2], [3], [7]}.