In 2024, Paragon released a major analysis of improper enrollment in the Affordable Care Act (ACA) health insurance exchanges. The research found that enhanced COVID-era subsidies and administrative actions by the Biden administration had allowed improper enrollment in the exchanges to run rampant, resulting in an estimated 4.84 million ineligible enrollees within the income group of people earning between 100 to 150 percent of the federal poverty line (FPL). The analysis focused on this group because enrollees in this income band received the largest subsidies, driving net premiums to zero and creating strong incentives for enrollees, unscrupulous brokers, and other dishonest insurance intermediaries, and health insurers to misreport income.
The paper’s research design was transparent and based on publicly available data. To estimate the numbers of ineligible enrollees, the authors (Brian Blase and Drew Gonshorowski) compared information from the Centers for Medicare and Medicaid Services (CMS) on the number of ACA exchange sign-ups who claimed income between 100 and 150 percent of FPL in 2024 to the estimated number of people with actual income in this range who would be eligible for such coverage.
To determine the size of the population eligible for coverage, the authors used the American Community Survey (ACS), the largest annual survey of the US population carried out by the Census Bureau. The ACS collects information from about 3.5 million households and produces highly reliable state-level estimates that are widely used in health policy research. Households surveyed by the ACS are asked a range of questions about their social, demographic, and economic circumstances. Using this detailed information, we derived estimates of the number of residents in the 100-150 percent of FPL group in each state who were eligible for ACA plans.
Due to its scale and complexity, the Census Bureau releases the results of the ACS with significant time lags. At the time Paragon’s report was prepared in early 2024, the most recent available year of ACS data was 2022. To improve comparability with 2024 data on enrollment in the ACA exchanges, the analysis incorporated a small adjustment to account for overall population growth from 2022 to 2024. In particular, the authors calculated the number of people eligible for exchange enrollment in each state in 2022 (using 2022 ACS data) and then, as stated in the paper, applied “population growth trends by state from 2020 to 2023 in order to approximate the number of people in income groupings in 2024.”
Since the number of people within income categories at the state-level changes slowly over time, this method provided a reasonable estimate of the eligible exchange enrollment in 2024. However, the authors cautioned that “this approach will not capture distributional changes that might be present” and acknowledged that “[the population] adjustment might not fully account for changes in distribution by FPL by state.”
Some critics of Paragon’s work, such as Keep Americans Covered, asserted that the use of 2022 ACS data – and the population adjustment to obtain estimates for 2024 – invalidated the study’s conclusions. Keep Americans Covered issued a critique of Paragon’s work that included the following statement:
“In addition to different data sources, Paragon’s analysis uses data from different years to produce misleading results. The analysis compares 2024 public use file enrollment data on the number of Marketplace plan selections to 2022 Census data, which was used to estimate the number of enrollees eligible for Marketplace coverage with incomes between 100–150% FPL. Comparing data from different years produces false results…”
Keep Americans Covered presented no quantitative evidence that the use of 2022 ACS data substantially skewed Paragon’s results. Moreover, re-analysis using the latest ACS data reveals that the attacks from Keep Americans Covered were spurious.
The Census Bureau has now released 2024 ACS data, allowing us to reassess the extent of ACA improper enrollment in 2024 without using prior-year data or incorporating any adjustment for population changes. We are now able to directly compare the number of eligible exchange enrollees using 2024 ACS data to the number of ACA exchange sign-ups who claimed income between 100 and 150 percent of FPL in 2024.
Using 2024 ACS data, we have also updated our 2025 estimates of improper ACA enrollment. In June 2025, Paragon released an analysis using 2023 ACS data, which required population adjustments due to data lags. The authors (Brian Blase, Chris Medrano, Niklas Kleinworth and Jackson Hammond) used the state-level growth rate of the overall population from 2023 to 2024 to convert the 2023 ACS figures to 2024. They then converted the 2024 estimates to 2025 by extrapolating the average annual state-level population change from 2020 to 2024. This extrapolation was necessary because the Census Bureau had not released 2024 to 2025 state-level population growth data at the time of the study’s publication. Using 2024 ACS data, we are now able to re-assess the extent of improper enrollment in 2025 with a single-year population adjustment (from 2024 to 2025), rather than the two-year adjustment implemented in the original study.
This updated analysis reveals that improper enrollment in the exchanges in both 2024 and 2025 was even more pervasive than Paragon’s original analysis suggested. State-level results are shown in Table 1. While the original 2024 analysis had found an estimated 4.84 million improper enrollees in the 48 states included in the study, the revised methodology using 2024 ACS data implies 5.11 million improper enrollees – a difference of nearly 275,000 enrollees, or about 6 percent. Similarly, the updated 2025 estimates indicate 6.47 million improper enrollees in 2025; our original analysis had identified 6.37 million improper enrollees in 2025 – a difference of 96,000 enrollees, or about 2 percent. Importantly, our methodology is conservative and likely undercounts actual improper enrollment because we assume that all eligible enrollees are enrolled in states where estimated improper enrollment exceeds the eligible population, and we assume that there is no improper enrollment in states that do not meet our threshold for what is considered improper enrollment.
