Will anyone be able to figure out American health care? So far, perhaps the world’s most byzantine arrangement of doctors, hospitals, clinics, contractors, pharmaceutical companies, private insurers, public insurers, medical schools, nursing homes, and dozens of other stakeholders has been less a coherent system than a collection of discount-furniture bits and pieces thrown on a floor with no instructions for assembly. Each individual piece usually works well and America’s doctors especially do pretty good jobs—that’s why they earn the big bucks—but fusing these disparate components to make a coherent health economy has often looked more like alchemy than science.

The Affordable Care Act has been the most recent attempt at transmuting the pieces of health care into a well-functioning whole. Recent news, however, including Aetna’s sudden exit from states’ health-insurance exchanges and forecasts of a spike in insurance premiums, has cast serious doubt on the chances of that undertaking succeeding. Is this turbulence to be expected or is it a sign that Obamacare is buckling under the strain of impossibility?

American health-care reform has always struggled to align two concepts that tend to be inversely related: access and affordability. Care is expensive to provide, but it doesn’t quite adhere to classic supply and demand curves for a number of reasons, including the fact that health insurance shields most patients from direct costs and because the government is so heavily involved in the market. Insurance is usually a good thing for patients, though, because it is the only thing that allows many Americans to afford even some basic health services without going bankrupt.

Insurance is, however, a major contributor to the irreconcilability of access and affordability. In most insurance markets, the incentive is for insurers to pay for as few things as possible in exchange for regular, guaranteed premiums; the risk insurers shoulder of having to pay for services for any enrollee is reflected in those premiums. Life insurance’s example as perhaps the ideal insurance market makes that clear: Everybody will die, of course, but death is an increasingly likely event for older people and people with certain conditions and behaviors, like smoking. No life insurer would take on an enrollee who is obviously already dying. So customers are charged more in their premiums—or denied insurance completely—based on their own risk factors. That’s how insurers keep the lights on.

American health policy, however, has generally steered the health-insurance market away from denying vulnerable patients coverage. Health insurance itself did not arise in an ideal market, since soon after its inception it became a recruiting tool for employers. The wide expectation of health insurance in the working class, and a developing health-care system that became focused on providing preventative and primary-care services, meant that insurance had to cover everyday services beyond the catastrophic events for which the classic insurance model is best suited. Tax incentives in the 1940s and 1950s made employer-sponsored insurance—in which employers and employees often split the costs of risk—essentially the backbone of American health care and provided affordable services for much of the middle class.

That development allowed insurers to tap into massive, stable populations of healthy adults who were backed by the stability of their employers’ contributions, and the resulting windfalls helped create the modern American system, which was premised mostly on a sprawling collection of doctors and hospital that rely on employees and retirees’ rich benefits to offset losses from sicker, poorer, and uninsured patients. From the ‘50s on, most of the money-makers in insurance risk pools had already been covered.

Each stage of health-care reform in the U.S. since then has involved a significant investment of public tax money to bend access and affordability closer and closer to meeting, while keeping the basic premium-based model in place. The creation of Medicare and Medicaid in 1965 provided government-sponsored medical safety nets to three of the riskiest groups: elderly people, people with disabilities, and poor families with children. Then, the Emergency Medical Treatment and Labor Act of 1986 expanded the authority of Medicare by stipulating that any hospital that accepted its patients also had to stabilize and treat any patient suffering a medical emergency, regardless of their ability to pay. The Children’s Health Insurance Program (commonly known as CHIP) in 1997 significantly expanded Medicaid’s pool to low-income families with children.

The main problem with that approach? It’s expensive for the government, which takes on the costs of risks and bloated health-care expenses. The economic argument for universal coverage is that covering everyone will provide a healthy mix of sick people and healthy people to balance risk and will promote the use of cheaper preventative care that eases the need for more expensive treatments later on. But those are downstream goals with considerable up-front price tags. The ACA is premised on that economic argument, originally seeking universal coverage by bolstering private insurance through state-run exchanges and employers, extending Medicaid eligibility to healthy low-income adults without children, providing subsidies for anyone left out, and compelling people to purchase insurance and employers to provide it. Each of those steps required complicated tax-code revisions and often shifted the costs of risk to the federal government, which was originally expected to have to set aside over $130 billion to cover those changes.

So far, the parts of Obamacare that have been the hardest to implement are the state insurance exchanges, in which people without affordable employer insurance or public coverage can shop for tiered, often-subsidized plans and the individual mandate, which requires them to do so. The ongoing issues with both show just how difficult reform can be. The creation of a robust self-purchase insurance market was integral to providing insurance to these people. But states and the federal government took turns making errors with these exchanges. Some states simply refused to cooperate, forcing the federal government to foot the bill and put in the time to create exchanges that could cater to local populations. Other states, like Oregon, struggled so much that federal administrators had to step in anyway. For its part, the federal rollout of the HealthCare.gov sign-up portal was an absolute disaster, and sign-ups continued to lag for years.

That’s where Aetna’s withdrawal from the insurance exchanges comes in. The states’ health-insurance exchanges are a rather small piece of the insurance pie—only about 12.7 million people had signed up for exchange plans as of the latest open enrollment period—and are dwarfed by the behemoth of employer coverage. But the people targeted by these markets are the most enigmatic and difficult for insurers to cover: Often they have too much income to qualify for Medicaid, yet are underemployed or below middle-class; they’re mostly a mix of “young invincibles” who are generally healthy and don’t see a high need for insurance and a group of older, sicker workers who are either unemployed or work in smaller firms and part-time jobs who need insurance but can’t afford it and aren’t yet eligible for Medicare. In order to incentivize this group to enroll, the ACA relies on a carrot and a stick. As the carrot, the ACA provides robust subsidies based on the exchange’s plan prices and income for people above Medicaid eligibility and making less than four times the poverty line. The stick is a tax mandate to purchase insurance.

So far, the carrot has been much more effective than the stick, and that’s not a good thing for the markets. The sickest older eligible people generally signed up for heavily subsidized health insurance, while the healthy younger people have been reluctant until recently, and new rules prohibiting insurers from denying coverage or adjusting premiums based on certain elements of risk have meant that they either take more losses, raise premiums, or do both if their patients turn out sicker than expected. It appears both are happening in tandem, and premiums for the exchange benchmark plans will rise by almost 10 percent on average this year as a result. That won’t be a big problem for most people in the market, as over 80 percent of all enrollees don’t actually see the true costs of insurance because of federal tax credits applied to premiums and cost-sharing of deductibles, copays, and coinsurance between patients and the federal government, but it is a big problem for the government and for insurers themselves.

In an attempt to control premiums and avert a “death spiral,” where rising costs and patient risk both continually intensify each other, the ACA also provides reinsurance, risk adjustment, and “risk corridors” to compensate insurers who accept sicker patients and experience higher costs than expected. In essence, these programs spread the gains of the whole individual market and of insurers that took lower-risk patients to mitigate the losses of those with sicker enrollees or higher costs than expected. Those programs should have functioned as stopgaps, temporarily encouraging individual-exchange insurers to pick up their fair share of sicker enrollees. And while these measures have funneled billions to insurers that have taken on losses to enter the market, risk corridors and reinsurance will be phased out in 2017, and risk corridors have been so far behind on payments that insurers launched a class-action lawsuit in February to seek compensation.

It’s no wonder, then, that Aetna suffered losses of $430 million since its entry into the exchanges. Like UnitedHealth Group before it, Aetna cited issues with the risk pool—that sicker patients are signing up more than healthy patients—in its decision to leave all but a handful of exchange markets. While that rationale is certainly suspect given the release of documents suggesting Aetna pulled out of the markets in retaliation for the Department of Justice blocking a merger with Humana, the move has an undeniable financial logic behind it, especially for an insurer of Aetna’s size. Why participate in a struggling, costly individual market when the lucrative honeypots of employer plans, privately-administered Medicaid plans, and Medicare advantage are there for the taking?

While Aetna’s move does highlight major issues in the exchanges, it probably isn’t a catastrophe for Obamacare. Kevin Counihan, the CEO of the federal insurance marketplace, expressed confidence in the markets after the move and in a blog post noted that the exchange risk pools are also “gaining healthier, lower-cost consumers” in the long-term. Government subsidies do at least help stabilize the market, so adverse selection won’t likely lead to the dreaded “death spiral” of ever-increasing costs and ever-sickening patient bases. Since essential covered benefits are standardized under the ACA, plans can compete by lowering the costs and increasing the efficiency of the services they provide.

Early returns from profitable insurers indicate this is happening. While the exchanges suffered losses of almost $3 billion in 2014 and were on track for heavier losses through 2015, a McKinsey report found that “carriers earning a positive margin in 2014 appear to share several common factors, including narrowed networks and managed plan design.” Kaiser HMO plans appeared to be major beneficiaries of that market preference, and several Blue Cross plans have jettisoned some less restrictive options for managed care. In the wake of Aetna’s exit, Blue Cross plans actually expressed confidence in their ability to manage care and even expand into new states. Given time, there is evidence that exchange markets will self-correct and provide a few models that successfully draw in balanced risk pools and minimize adverse selection. That self-correction would, however, inevitably result in more high-profile insurer exits like Aetna’s—which, in that sense, was actually part of the plan.

Health-care reformers can’t afford to wait for that self-correction because plan exits like Aetna’s put people at the mercy of an inherently volatile environment and run the risk of violating Obama’s central pitch about keeping plans. They also run the risk of making Obamacare easy political fodder for Republican campaigners. Pinal County, Arizona, might be left without any exchange insurers after Aetna’s withdrawal next year, and roughly a quarter of all counties in the country are already left with only one option. While most people won’t feel the effects of plan exits until next year and don’t shoulder the burden of premium increases, premiums have risen for many families and any dysfunctions in the controversial reforms are easy political targets. Counihan has signaled that exchanges will aggressively recruit more insurers for 2017 to ameliorate the attrition. He has also suggested that the administration will fine-tune its risk-adjustment strategy to better deal with high-risk patients and will open doors for states like Alaska to address specific market needs with program waivers.

Those are fairly minor tweaks, though, and even several champions of the 2010 law are pushing for major changes to health law. The public option seems to have gotten the most traction after Aetna’s exit, and it still polls favorably among most Americans, despite being excluded from the original health-reform debate early on. Such an option could be administered by private insurers, states, nonprofits, or a mix across the states would be backstopped by tax dollars, and would guarantee the existence of at least one market-proof option in every county and state, all the while siphoning off some of the riskiest components of the exchange pools. Proponents have cited the size and bargaining power of the federal government, as well as the resulting efficiency and market competition, when making the case for the public option. It is still unclear if that theorized bargaining power and efficiency would be able to fix the risk problem in the exchange markets, however.

One possible fix to those risk pools could be simply restoring the original form and function of the ACA. Expanded Medicaid was originally intended to use federal and state funds to cover all people under 138 percent of the federal poverty line, but the Supreme Court decision in NFIB v. Sebelius gave states an opt-out for that expansion. Nineteen states have chosen not to expand Medicaid, and in those states, the floor for exchange subsidies is lowered to the poverty line, below which people are not mandated to purchase insurance. This places about two million people who would be eligible for Medicaid into exchange markets, and as is roughly true generally, these lower-income participants are more likely to be sicker than participants in the intended risk pools. Of the 18 states that had the worst exchange performance among insurers in the McKinsey report, 12 have not yet expanded Medicaid to all low-income adults and three expanded the program after the exchanges went live. One of the remaining three was Oregon, where the $200 million debacle of the health-insurance-exchange implementation may have had long-term effects on sign-ups of healthy people. A choice by states to expand Medicaid as intended, perhaps in combination with a plan such as Hillary Clinton’s “Medicare for more” plan, which would extend Medicare buy-in options for people over 55, could help balance risk in the exchange markets and put more of the sickest patients into government-managed health care.

Of course, if the general strategy over 50 years of policy has basically been to shift as many sicker people and as many costs onto the government, why not just go all the way? Single-payer health care would solve the problem of the exchange market by merging it with the stable, lucrative public-insurance programs and employer-sponsored insurance. That approach would necessitate higher taxes, probably administered across wages, businesses payrolls, and the health-care industry, but those taxes would replace current premiums and employer contributions. That system would eliminate the fractious nature of the health-care system that obfuscates price and often makes competition meaningless or even occasionally increases costs. And in a best-case scenario, single-payer might present a single, massive and coherent entity to negotiate directly with powerful health-care providers for lower prices. It would be the clearest way of solving the access-affordability conundrum.

The biggest problem with single-payer (beyond the politics) is not increased taxes, but that people don’t really want to give up their plans or physicians. That’s why Obama’s original pitch about being able to keep health plans was important to public opinion and why Aetna’s threat of undermining that pitch is so damaging. The plurality of Americans approve of the government paying for health care, but that approval plummets when the proposal involves the government dictating which services they receive and from where. The solution to avoid total disruption might look something more like the German system, which is mostly publicly financed but privately administered. Individuals’ contributions—essentially income taxes—are collected by a central government funder and then divided up among municipal and employer-based nonprofit insurers that represent each worker and their families, with taxes and municipalities picking up the tab for recipients of welfare. The German system has its own problems—for one, it requires a high ratio of workers to retirees to keep the tax support going—but it is the universal public-funded option that most resembles the American hodgepodge.

Aetna’s withdrawal from most Obamacare markets does highlight the issues that make health-care reform so difficult, and it does show some of the deeper problems in the structure of the ACA. But the takeaway probably isn’t that a sudden, dramatic collapse is imminent, but that health-care reform is a process that has always strained to meet the same two goals of access and affordability. The best news for policymakers on either side of the political aisle is that there are still several policies that can help get there.