CBO’s March 2017 health bill score traced most early coverage losses to the mandate penalty, later zeroed
CBO’s March 13, 2017 estimate of the American Health Care Act tied the 2018 coverage loss mainly to ending the mandate penalty. The bill failed, but that penalty was set to zero in December 2017. Health systems were better served modeling provisions than bills.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
CBO’s March 13, 2017 estimate of the American Health Care Act tied the 2018 coverage loss mainly to ending the mandate penalty. The bill failed, but that penalty was set to zero in December 2017. Health systems were better served modeling provisions than bills.
- The week The coverage figures drew more attention.
- The penalty: the channel with the shortest fuse The agencies traced the bulk of the 2018 coverage loss to repealing the penalties. [1] Their reasoning had two strands.
- The market rules: the channel through insurer pricing The second channel worked through how insurers priced and who bought.
- The subsidy rebuild: the channel through household budgets The third channel was the replacement of the ACA’s subsidies.
- The Medicaid financing changes: the channel through state budgets The fourth channel ran through states.
- How the channels land on different providers The four channels reach providers through different doors.
- Objections, in order The bill did fail. [2] But its first channel did not.
- What finance and revenue cycle leaders could check The four channels translate into checks for any health system facing a federal coverage bill, then or now.
- With hindsight The American Health Care Act did not become law.
The week
On March 13, 2017, the Congressional Budget Office and the staff of the Joint Committee on Taxation published their cost estimate for the American Health Care Act. [1] The estimate covered the budget reconciliation recommendations that the House Committees on Ways and Means and Energy and Commerce had approved, in a version dated March 9, 2017. [1] The fiscal year 2017 budget resolution had directed those two committees to develop legislation to reduce the deficit. [1]
The headline budget figure was a reduction in federal deficits of $337 billion over the 2017–2026 period. [1] Outlays would fall by $1.2 trillion and revenues by $0.9 trillion over the same period. [1] By the agencies’ account, the largest savings came from lower Medicaid outlays and from ending the Affordable Care Act’s subsidies for nongroup coverage. [1] The largest costs, they said, came from repealing tax changes the ACA had made and from a new tax credit for health insurance. [1]
The coverage figures drew more attention. The agencies estimated that 14 million more people would be uninsured in 2018 than under current law. [1] They said the larger part of that increase would stem from repealing the penalties associated with the individual mandate. [1] The gap would rise to 21 million in 2020 and 24 million in 2026. [1] In 2026, an estimated 52 million people would be uninsured, against 28 million under current law. [1]
The estimate broke the 2018 change down by type of coverage. It described about 6 million fewer people with nongroup coverage, roughly 5 million fewer with Medicaid and about 2 million fewer with employment-based coverage. [1] The agencies explained that some people bought insurance under current law only to avoid the penalty, and others would drop coverage because premiums rose. [1]
The estimate listed the provisions that drove those effects. [1] The bill would eliminate the penalties tied to the individual and employer mandates. It would reduce the federal matching rate for adults made eligible for Medicaid by the ACA, beginning in 2020. It would cap growth in per-enrollee federal Medicaid payments from 2020. [1] It would repeal the ACA’s premium tax credits and cost-sharing subsidies from 2020 and create a new refundable tax credit. [1]
The market provisions came next on the list. The bill would fund grants to states through a Patient and State Stability Fund beginning in 2018. [1] It would let insurers charge older people five times as much as younger people, rather than three times, unless a state set a different limit. [1] It would remove, from 2020, the requirement to offer plans covering at least 60 percent of the cost of covered benefits. [1]
One provision replaced the mandate with a different incentive. Insurers would have to apply a 30 percent surcharge on premiums for up to 12 months to people enrolling after being uninsured for more than 63 days in the past year. [1] The agencies expected it to bring roughly 1 million people into coverage in 2018, then deter roughly 2 million a year after that. [1]
On premiums, the estimate projected that average nongroup premiums for single policyholders would be 15 to 20 percent higher than under current law in 2018 and 2019. [1] It attributed that mainly to the end of the mandate penalties. [1] By 2026, average premiums would be roughly 10 percent lower than under current law, partly because of a younger mix of enrollees. [1]
The estimate also touched hospitals directly. It projected a net increase of $43 billion in Medicare spending from 2018 to 2026, from changes in payments to hospitals that serve a disproportionate share of low-income patients. [1] Lower Medicaid enrollment would slightly reduce some of those payments, it said, but more uninsured people would substantially boost the amounts distributed from the lump-sum pool. [1]
The agencies called their estimates uncertain. They said the responses of agencies, states, insurers, employers, individuals, doctors and hospitals were difficult to predict, and that they had aimed for the middle of the distribution of outcomes. [1] Because of the short time available, they had not quantified the bill’s macroeconomic effects. [1] They assumed enactment by May 2017. [1]
The March 13 estimate is remembered for its coverage totals. We read it for something less noticed: it said when each loss would arrive and which provision would cause it. The 2018 losses, it said, came mainly from ending the mandate penalty. [1] The later losses came from subsidy and Medicaid changes that would start in 2020. [1]
That distinction became important within the year. The bill itself did not become law. The House postponed consideration on March 24, passed a later version on May 4, and the Senate did not pass it. [2] Yet in December 2017, Congress enacted a separate law that set the individual mandate penalty to zero for months after December 31, 2018. [4]
Our argument is that the estimate was a map of separable provisions, each on its own clock, and that health systems were better served by reading it that way than by tracking the bill. The provision that mattered first in the estimate was also the one that survived the bill’s failure. A finance team that modeled provisions rather than bills would have had the first piece already scored.
We take the estimate’s coverage effects in four parts. Each is a different channel with its own start date:
- The penalty: ending the individual mandate penalty, effective at once in the bill’s design.
- The market rules: age rating, continuous coverage and stability grants, from 2018 and 2019.
- The subsidy rebuild: replacing ACA subsidies with a new tax credit in 2020.
- The Medicaid financing changes: the matching rate and per-enrollee caps from 2020.
The first part acts through individual choice. The second acts through insurer pricing. The third acts through household budgets. The fourth acts through state budgets. After them we look at how the channels land on different providers, take the objections in order, and end with checks a health system could run.
The penalty: the channel with the shortest fuse
What the estimate said
The agencies traced the bulk of the 2018 coverage loss to repealing the penalties. [1] Their reasoning had two strands. Some people held coverage under current law only to avoid the penalty, and would drop it. [1] Others would drop it because premiums rose as healthier people left the market. [1]
The estimate put numbers on that second strand. Nongroup premiums for single policyholders would run 15 to 20 percent above current law in 2018 and 2019, mainly because the penalties would end. [1] It expected nongroup enrollment in 2018 to be smaller than in 2017, with worse average health among those who stayed. [1]
Even 2017 was in scope
The estimate did not wait for 2018. It said the coverage increase would be limited at first, because insurers had already set premiums for 2017 and many people had made their enrollment choices. [1] Even so, it estimated that eliminating the penalties would leave about 4 million additional people uninsured in 2017. [1] In 2019, the gap would grow to 16 million, through further Medicaid and nongroup reductions. [1]
The agencies gave totals for the population under 65. They estimated 41 million uninsured in that group in 2018 and 43 million in 2019 under the bill. [1] Those figures set the base against which a health system could size its own share.
How the agencies modeled the choice
The estimate explained its method. CBO used a microsimulation model built on survey data about a representative sample of individuals and families, including income, employment, health status and coverage. [1] The model drew on research about how people and employers respond to price changes and to changes in eligibility for public coverage. [1]
That method has a consequence for anyone translating the results. The model works on the population in aggregate. It does not know a hospital’s catchment area, its payer contracts or the employers in its town. A health system that wants to apply the estimate has to rebuild the chain on its own population. It needs to know who holds which coverage, how price-sensitive they are, and who was buying coverage because of the penalty.
Why it reached Medicaid and employer coverage too
The penalty channel was not confined to the exchanges. The 2018 decline included roughly 5 million fewer people on Medicaid and about 2 million fewer with employment-based coverage. [1] In 2018, the bill’s Medicaid financing changes had not yet begun. The estimate explained the link. Under current law, the penalties applied to some Medicaid-eligible adults and children, such as single people with income above about 90 percent of the poverty guidelines. [1] Without them, CBO expected fewer people to enroll in Medicaid, including some who were not subject to the penalties but might think they were. [1]
That detail deserves a second look. It says a penalty can drive enrollment among people it does not legally reach, through uncertainty about income or the rules. [1] Removing it then reduces enrollment among the same people. For a hospital, the patients affected may be ones it would have screened as Medicaid-eligible at registration.
That matters for hospitals. A patient who drops Medicaid coverage still arrives at the emergency department. A health system’s self-pay volume and charity care depend on how many patients arrive without coverage, whatever the reason they lack it. The estimate put that change in 2018, before any Medicaid formula changed. [1]
The employer requirement
The bill also ended the penalty tied to the employer requirement, while keeping the requirement itself. [1] That requirement was that large employers offer coverage meeting specified standards. [1] The agencies said eliminating both penalties would reduce federal revenues from 2017 and would substantially reduce the number of people with coverage. [1] They did not separate the employer penalty’s share of the roughly 2 million fewer people with employment-based coverage in 2018. [1]
The estimate gave the logic for the individual side. Insurers would still have to accept every applicant, could not price on health status, and would be limited in pricing by age. [1] Those features attract applicants with higher expected costs. So repealing the penalty would reduce coverage less among older and less healthy people than among younger and healthier ones. [1] In the agencies’ words, the repeal taken by itself would increase premiums. [1]
What a finance team could have done with it
The penalty channel had a near-term date and a clear mechanism. A health system could test its own payer mix against it. Which service lines drew patients from the nongroup market? Which drew younger adults likely to drop coverage once the penalty ended? Those questions could be answered from the system’s own data, without waiting for Congress.
The answer would differ by market. A system in a state with a large exchange population faced more exposure through this channel than one where employer coverage dominated. The estimate gave national figures, not state ones. [1] Translating them was the health system’s job.
The market rules: the channel through insurer pricing
Age bands and the stability fund
The second channel worked through how insurers priced and who bought. The bill would let insurers charge older enrollees five times as much as younger ones, unless a state chose otherwise. [1] The estimate’s list of provisions dated the new limit from 2018, but its premium analysis treated the age-rating change as effective in 2019. [1] It would cut premiums substantially for young adults and raise them substantially for older people. [1]
The Patient and State Stability Fund would pay out $80 billion over 2018–2026, by the agencies’ estimate. [1] They expected states to use the grants mainly to reimburse insurers for enrollees with claims above a threshold. [1] Before 2020, they expected the federal government to run a default program for most states, which lacked time to set up their own. [1]
Plans with less coverage
The market channel included a change to plan design. Under current law, plans in the nongroup and small-group markets generally had to fall in one of four tiers of actuarial value: about 60, 70, 80 or 90 percent. [1] Beginning in 2020, the bill would repeal those requirements, potentially allowing plans below 60 percent. [1] Other protections would stay, including required categories of services and limits on out-of-pocket costs. [1]
For providers, the actuarial value of a patient’s plan determines how much of a bill the patient owes. A plan that pays a smaller share of costs leaves more to collect from the patient. In our reading, this provision would have moved some revenue from insurer payments to patient balances, which are harder to collect. It would show up as bad debt rather than as uninsured visits.
Continuous coverage
The surcharge provision replaced the penalty with a price on gaps. A person uninsured for more than 63 days in the past year would pay a 30 percent premium surcharge for up to 12 months on enrolling. [1] It would apply to special enrollment in 2018 and to all enrollment from 2019. [1]
The agencies expected it to work backward. It would pull roughly 1 million people into coverage in 2018 to avoid a future surcharge. [1] After that, roughly 2 million fewer a year would buy coverage because of the surcharge or the paperwork to prove prior coverage. [1] The agencies expected those deterred to be healthier than those who paid. [1]
The agencies’ stability judgment
The estimate addressed whether the market would hold. In the agencies’ assessment, the nongroup market would probably be stable in most areas under either current law or the bill. [1] They gave three reasons under the bill: subsidies would keep enough relatively healthy people buying, stability grants would reduce insurers’ losses on very large claims, and risk adjustment would remain. [1]
They also flagged a weakness. Significant changes in subsidies and market rules would occur in each of the first three years after enactment, which might make premiums hard to set. [1] In 2019, a one-year change to the premium tax credits and the new age rule would shift the mix toward younger adults. [1] In 2020, new credits and the end of actuarial value tiers would arrive together. [1]
For a provider, year-by-year change in insurer pricing has a direct cost. Contracts with insurers are negotiated on expectations about membership and mix. When insurers cannot predict their own enrollees, the provider’s contract assumptions become guesses as well. In our reading, the estimate’s warning about premium setting was also a warning about payer contracting.
What this channel meant for providers
For providers, the market channel changed who in a community held coverage, as well as how many. Older adults in the nongroup market faced higher premiums under the age rule. [1] The agencies expected the increase in uninsured to be disproportionately larger among people aged 50 to 64 with income below 200 percent of the poverty line. [1]
In our reading, a shift of uninsurance toward older, lower-income adults could weigh more heavily on hospital charity care than the same headcount among young adults. The estimate did not quantify that effect for providers. A health system could.
The subsidy rebuild: the channel through household budgets
From income-based to age-based credits
The third channel was the replacement of the ACA’s subsidies. Under current law, premium tax credits went to people with income between 100 and 400 percent of the poverty line, with exceptions, and limited what they paid as a share of income. [1] The bill would repeal them and the cost-sharing subsidies in 2020. [1] A new refundable tax credit would take their place. [1]
The agencies said the new credits would generally be less generous for those receiving subsidies under current law. [1] They still judged that the nongroup market would probably be stable in most areas, because of the credits, the stability grants and a younger mix of enrollees. [1] The new credit would cost $361 billion over the period, against savings of $673 billion, mostly from ending the old subsidies. [1]
The shape of the new credit
The estimate described the new credit in detail. It would vary by age by a factor of 2 to 1, from $2,000 for someone under 30 to $4,000 for someone 60 or older. [1] People would generally get the full amount with adjusted gross income below $75,000 for a single filer or $150,000 for joint filers, with a phase-out above. [1] Credits and thresholds would be indexed to consumer prices plus 1 percentage point. [1]
The contrast with current law was in what moved the subsidy. The ACA credit depended on household income and the reference premium in the enrollee’s area, which insulated the enrollee from local premium variation. [1] The new credit would not track local premiums. In our reading, that made the subsidy channel geographic: enrollees in high-cost areas would feel the change more than those in low-cost ones.
That geography maps onto providers. High premiums in a rating area reflect, among other things, the prices of local hospitals and physicians. A credit that ignores local premiums would leave enrollees in expensive markets with larger shares to pay. Health systems in those markets would face more pressure on coverage among their patients.
An administrative dependency
The estimate also named an operational assumption. Because the new credits were designed to be paid in advance to insurers, the IRS and HHS would need procedures to verify that payments went to eligible insurers on behalf of eligible enrollees. [1] The estimate assumed adequate resources would be appropriated for those systems in time. [1] If they were not, it said, enrollment and compliance could be affected. [1]
That kind of assumption deserves attention from any organization modeling a federal program change. A provision on paper can be delayed by the systems needed to run it. Health systems that depend on federal subsidies reaching their patients carry that implementation risk too.
Employers’ responses
The estimate also considered employers. Some might stop offering coverage and raise other compensation, judging nongroup insurance a close substitute. [1] The agencies expected employers that dropped coverage to have younger and higher-income workforces. [1] They expected employers to adapt slowly, with some delaying decisions because of uncertainty. [1]
For a health system as an employer, that was a second exposure. Systems buy coverage for their own staff. A change in the nongroup market could affect their benefits strategy as well as their patients. The two effects would run through different budgets inside the same organization.
Timing
The rebuild would start in 2020. [1] That gave the channel a longer fuse than the penalty. It also meant more chances for the policy to change before it took effect. A provision with a later start date carries more political risk, because later Congresses can amend it. In our reading, a careful planner would weight a 2020 provision below a 2018 one even if the bill passed.
The Medicaid financing changes: the channel through state budgets
What the bill proposed
The fourth channel ran through states. From 2020, the bill would reduce the federal matching rate for the ACA expansion population to each state’s regular rate. [1] It would cap growth in federal per-enrollee payments for most beneficiaries at the medical care component of the consumer price index. [1] The agencies estimated an $880 billion reduction in federal Medicaid outlays over 2017–2026. [1]
The estimate expected states to respond. With states paying a greater share, the agencies expected no additional states to expand eligibility. [1] They expected some states that had already expanded to stop offering that coverage. [1] By 2026, they estimated, about 30 percent of the newly eligible population would live in a state with expanded eligibility. [1]
The grandfathering mechanism
The matching rate change had a built-in phase. The lower rate would apply only to people newly enrolled after December 31, 2019, or who had a break in eligibility after that date. [1] The savings would therefore grow over several years as earlier enrollees cycled off. [1] On the basis of historical data, CBO projected that fewer than one-third of those enrolled at the end of 2019 would keep continuous eligibility two years later. [1]
The estimate noted that the bill required more frequent eligibility redeterminations. [1] It projected that the higher matching rate would apply to fewer than 5 percent of newly eligible enrollees by the end of 2024. [1] In our reading, that design turned eligibility churn into a financing lever. Each redetermination would be a point at which a state’s cost for an enrollee could rise.
For hospitals, churn is an operational matter as well as a financial one. A patient who loses Medicaid eligibility between visits arrives as self-pay. Eligibility and enrollment teams, financial counselors and presumptive eligibility processes become the front line. The estimate’s mechanism would have put more weight on those teams over time.
How the per-enrollee cap worked
The cap set a limit on federal reimbursement from 2020. [1] It would start from each state’s average per-enrollee cost in 2016 for five groups: the elderly, disabled people, children, newly eligible adults and all other adults. [1] Those averages would grow with the medical care component of the consumer price index. [1] Spending above the limit would fall to the state. [1]
The base year matters. A state’s 2016 spending pattern would become its permanent starting point. In our reading, that is a decision about the past that binds the future, and states with low 2016 spending per enrollee would carry that position forward.
Why this channel is separate
The Medicaid channel did not act through individuals’ choices, as the penalty did. It acted through state legislatures and budget offices deciding what to cover. Its timing would depend on state calendars as well as federal ones. A hospital association’s state budget team, not its federal team, would see it coming first.
We keep this part brief. Its later history in the Senate belongs to another week. For this week, the point is that the estimate assigned them to 2020 and later, and to a decision-maker other than the patient. [1]
Medicare disproportionate share payments
One provision cut across the channels. The agencies expected Medicare spending to rise by $43 billion over 2018–2026 because of changes in disproportionate share hospital payments. [1] Those payments have a per-case component and a lump-sum pool allocated by each qualifying hospital’s share of certain days of care. [1] More uninsured people would substantially increase the lump-sum amounts, the estimate said. [1]
In our reading, that line shows how far the estimate tracked effects into provider payment. It also showed that a rise in uninsurance would not hit every hospital the same way. Qualifying hospitals would see some offset through Medicare. Others would see none.
The offset also had its own timing and mechanics. The lump-sum pool is allocated by formula, so a hospital’s share depends on data about its own patient days. [1] Finance teams at qualifying hospitals would need to forecast both their uncompensated care and their share of a pool that grows with national uninsurance. Those two forecasts move together but not in step. A hospital whose own uninsured volume rose faster than the national figure would see the offset lag its losses. One whose volume rose more slowly could see the reverse. The estimate gave the national direction, not any hospital’s share. [1]
How the channels land on different providers
The four channels reach providers through different doors. The weight of each depends on a provider’s patients, location and structure. We set out the differences as analysis, not as findings of the estimate.
Health systems with large emergency volumes
Systems with high emergency department volume are likely to see uninsured patients first. The penalty channel, with its 2018 start, reached them soonest. [1] Their revenue cycle teams could track self-pay registrations by month. A change in that series would be an early signal that the estimate’s first channel was operating.
Academic medical centers
Academic centers draw patients from wide areas and treat complex cases. Their exposure depends on referral patterns and payer contracts. Older, sicker patients in the nongroup market were where the age-rating change concentrated premium increases. [1] For an academic center, those patients can carry large bills.
Community and rural hospitals
A smaller hospital may depend on a few payers. A rural hospital in a state that would stop its Medicaid expansion faced the fourth channel directly. One in a non-expansion state faced it less. The estimate expected some expansion states to stop offering that coverage, without predicting which. [1]
Ambulatory networks and physician groups
Physician practices and ambulatory networks see insured patients by appointment. If coverage lapses, some patients may defer visits. For these organizations, the penalty and market channels show up as lower volume and higher bad debt rather than as emergency charity care. Their planning horizon is shorter than a hospital’s.
Health systems that operate health plans
Some systems own insurers selling in the nongroup market. For them, the market channel is a pricing problem. The estimate expected premiums to rise in 2018 and 2019 and to fall below current law by 2026. [1] It warned that changes in each of the first three years might make premiums hard to set. [1] An integrated system would face that uncertainty on both its insurance and its provider books.
Health systems as employers
Every channel also reached health systems through their own payrolls. A system that offers coverage to staff faces the same employer decisions the estimate modeled. [1] The estimate expected employers that dropped coverage to have younger and higher-income workforces. [1] Health systems employ staff across a wide range of ages and incomes, so the model would not map neatly onto them.
The employer question links to the patient question. Staff who lose employer coverage may become patients without coverage at their own hospital. A benefits decision made in human resources can therefore appear later in the charity care budget.
Public and safety-net hospitals
Public hospitals answer to state and local budgets as well as to patients. They meet the fourth channel twice: through Medicaid revenue and through the government that funds their deficits. For them, a state budget office’s response to lower federal matching would matter as much as the coverage numbers.
Building the provision model
The analysis above suggests a practical build. A provision model needs four inputs for each channel: the start date, the population reached, the system’s current exposure to that population, and an assumed response. The estimate supplies the first two for each provision. [1] The health system supplies the third from its own registration and claims data. The fourth is a judgment, ideally tested against a range. A low, central and high response for each provision is enough to show which assumptions drive the result. If the central case changes the budget and the high case changes staffing, the model has told management where to look first.
The outputs should be operational as well as financial. A rise in self-pay volume has staffing consequences for financial counseling. A shift in age mix among the insured has consequences for service line planning. A change in Medicaid churn has consequences for eligibility teams. A model that ends in a single revenue number hides the work that follows.
Objections, in order
1. The bill failed, so the estimate did not matter
The bill did fail. [2] But its first channel did not. The December 2017 law set the penalty amount to zero for months after December 31, 2018. [4] A planner who had treated the estimate as a scoring of separable provisions would have found the penalty channel already modeled. Planning for a failed bill was wasted effort only for those who planned for the bill as a whole.
2. Coverage estimates are too uncertain to plan on
The agencies themselves said the estimates were uncertain. [1] Uncertainty argues for ranges and triggers, not for ignoring the mechanism. The estimate’s value for a health system lay less in the totals than in the causal chain: which provision, which population, which year. A system can watch its own data to see whether the chain is operating.
3. The penalty was small, so ending it could not matter much
One objection held that the penalty was too weak to change behavior. The agencies disagreed in March 2017, attributing the larger part of the early coverage loss to its repeal. [1] In November 2017, they estimated that repealing the mandate alone would leave 4 million more people uninsured in 2019 and 13 million more in 2027. [3] Those figures were lower than the March bill estimate’s totals, which included other provisions. The agencies added that preliminary work with revised methods pointed to effects that would probably be smaller still. [3] Even so, they said the number of uninsured people would be millions higher. [3]
4. Hospitals cannot influence federal legislation, so modeling is pointless
Modeling is not lobbying. A hospital models provisions to set budgets, reserves, staffing and charity care policy. Those decisions are inside its control. A model that separates provisions also lets a system respond quickly when one provision passes and the rest does not.
5. Health systems should plan on the most likely outcome only
Planning on one outcome left systems exposed in 2017. The outcome the March estimate assumed, enactment by May 2017, did not happen. [1] [2] A separate outcome, a zero penalty, did. [4] A planner who held scenarios by provision would have covered both. In our view, scenario planning by provision is the lesson of the year.
6. Medicaid matters more to hospitals than the individual market
For a safety-net hospital, Medicaid financing may be the larger exposure, and we treat it as a separate channel for that reason. But the estimate put the Medicaid financing changes in 2020 and later. [1] The penalty channel would act sooner, and its 2018 effects included Medicaid enrollment. [1] Both deserve a line in the model.
7. The later estimate was smaller, so March overstated the penalty
The November 2017 estimate projected 4 million more uninsured in 2019 from repealing the mandate alone. [3] The March estimate projected 14 million more in 2018 for the whole bill. [1] The two are not the same measure. The March total included the bill’s other provisions and was measured against CBO’s March 2016 baseline. The November figure used the summer 2017 baseline and assumed a slower phase-in of responses. [1] [3] Both estimates placed the penalty at the front of the timeline.
What finance and revenue cycle leaders could check
The four channels translate into checks for any health system facing a federal coverage bill, then or now.
- Have we broken the bill’s scored effects into provisions, each with its own start date?
- For each provision, do we know which of our patient populations it reaches first?
- Do we track a monthly signal, such as self-pay registrations, that would show the first channel operating?
- Have we modeled the state budget response separately from the federal provision?
- Do we know which provisions could pass alone, in another bill, if the main bill fails?
The checks follow the parts. The first and fifth check the argument: provisions, not bills. The second and third check the penalty and market channels. The fourth checks the Medicaid channel. The subsidy channel sits inside the first, as one provision with its own date.
Each check has an owner. Finance owns the provision model. Revenue cycle owns the monthly signal. Government relations owns the list of provisions that could travel in other bills. The state budget question belongs to whoever tracks the state legislature. Bringing those owners together for one meeting is the cheapest form of preparation available.
A record of the model belongs in the board file. It should show the assumptions for each provision, the date each was last reviewed, and the sources behind them. When a provision passes, fails or moves to another bill, the file should show the update. That record lets a finance committee see whether management tracked the right risks, and lets auditors trace a reserve or a budget change to its basis.
The record also protects against a less visible failure. A model built for one bill can be abandoned when the bill dies. In 2017, the provision that mattered first outlived its bill by a matter of months. [2] [4] A file that kept the provision alive after the bill failed would have been ready for the December law.
A short exercise makes the point. Take any scored federal health bill. Copy its provision list into a spreadsheet. Add, for each line, the start date, the population affected, the system’s exposure and the chance that the provision could pass in some other vehicle. The 2017 estimate would have filled that sheet with four distinct rows. [1]
With hindsight
The American Health Care Act did not become law. The House began debate on March 24, 2017, then postponed further consideration that afternoon. [2] It passed an amended version on May 4, 2017, by a recorded vote of 217 to 213. [2] On July 25, 2017, the Senate agreed to a motion to proceed to the bill. On July 28, 2017, the Senate rejected amendment 667 by 49 to 51, and the bill returned to the calendar. [2]
The first channel took another route. On November 8, 2017, CBO and JCT published an updated estimate of repealing the individual mandate by itself. [3] They estimated that repeal starting in 2019 would reduce deficits by about $338 billion over 2018–2027. [3] They estimated it would increase the number of uninsured by 4 million in 2019 and 13 million in 2027. [3]
The agencies added that eliminating the penalty without repealing the mandate would give very similar results. [3] With no penalty, they said, only a small number of people would keep buying coverage solely out of willingness to comply with the law. [3] They expected nongroup markets to remain stable in almost all areas, with premiums about 10 percent higher in most years. [3]
On December 22, 2017, the tax law known as Public Law 115-97 was approved. [4] Its section 11081 struck the 2.5 percent and $695 penalty amounts in section 5000A and inserted zero percent and $0. [4] The change applied to months beginning after December 31, 2018. [4] The 2023 edition of the United States Code shows section 5000A with the zero percent and $0 amounts, citing that section. [6]
The zero penalty then became the basis of a lawsuit. Texas, other states and two individuals argued that without a penalty the mandate was unconstitutional, and that the rest of the ACA could not stand without it. [5] A district court agreed on both points. [5] The Fifth Circuit agreed on standing and that the mandate was unconstitutional, but found the severability analysis insufficient and remanded. [5]
On June 17, 2021, the Supreme Court decided California v. Texas. [5] It held that the plaintiffs lacked standing to challenge the minimum coverage provision. [5] Without a penalty, the Court reasoned, the provision was unenforceable, and the plaintiffs had not shown an injury traceable to its enforcement. [5] It reversed the Fifth Circuit on standing, vacated the judgment and remanded with instructions to dismiss. [5]
What held up from March 2017 is the estimate’s causal map. The penalty channel was identified as the earliest driver of coverage loss. [1] It was the piece that became law. [4] In November 2017, the agencies again estimated that ending it alone would raise the number of uninsured, while noting that revised methods would probably give smaller effects. [3] The other three channels, as written in the House bill, did not become law in 2017. [2]
What did not hold up was the timing the estimate assumed. It assumed enactment by May 2017, with the penalty repeal reducing coverage from 2017 itself. [1] The law that actually ended the penalty took effect for months after December 31, 2018. [4] Health systems that tracked the bill’s date would have expected the change well before it came.
We do not report national uninsured rates after 2019 here. The sources we cite are estimates made before the change took effect, and the law and the Court decision that followed. A health system that wants to test the 2017 estimate against outcomes can compare its own self-pay and charity care data across 2018 and 2019. That comparison would be specific to its market, a level the national estimate did not address. [1]
As checked on November 15, 2026, the zero penalty in the 2023 edition of the Code is the codified text we cite. [6] The Supreme Court’s 2021 decision left the ACA standing without ruling on the mandate’s merits. [5] For health systems, the durable lesson of the week is about method: read a scored bill as a set of provisions with separate clocks, because Congress may enact them one at a time.
Spotted an error? Tell us and we will check it and correct the article where needed.