At its very core, an insurance company operates by: 1) pooling policyholder’s risks together and 2) collecting premiums from the policyholders based on actuarial data, to pay claims and maybe make a small profit. But looking broader, an insurance market exists when: a) policyholders voluntarily or are obliged to obtain policies, b) insurers are willing and able to accept the risks in exchange for a premium expected to support the insurance pool, and c) the actuarial risks are calculable and prove true, on average.

The loss of any of A, B, or C will substantially impact a healthy insurance market, or can prevent the insurance market from ever getting started. For some examples of market failures, the ongoing California homeowner insurance crisis shows how losing B (starting with insurers refusing to renew policies near the wildland-rural interface) and C (increase in insured losses due to climate change) results in policies becoming unaffordable or impossible to obtain.

As a broader nationwide example, an established business sector that operates wholly without insurance availability is cannabis. A majority of US States have decriminalized marijuana for medical use, and a near-majority have legalized recreational consumption. Yet due to unyielding federal law, no insurer will issue business policies, to protect from risks that all businesses would face, such as losses from fire, due to a product recall or product liability, or for liability to employees. These risks are calculable and there’s a clear need for such policies – thus meeting criteria A and C – but no commercial insurer is willing to issue. Accordingly, the formal market for cannabis business insurance is virtually non-existent in the USA.

With these examples, we can see that the automobile insurance market meets all three criteria for a healthy market, but it’s how these criteria are met which is noteworthy. Motorists in the USA are obliged in every state except New Hampshire and Virginia, it is a criminal offense to drive a car without third-party liability insurance, meaning the motorist might spend time in jail. Note: NH and VA won’t send a motorist to jail, but they do have administrative penalties for driving without “financial responsibility”, which includes insurance or a bond at the DMV.

The exact requirement varies per state, with some requiring very low amounts of coverage and others requiring extra coverage a like Personal Injury Protection (PIP, aka no-fault insurance). The point is that criteria A is easily met: motorists want to avoid jail, but also want to avoid the indignity of being sued after having caused a road incident, in addition to protecting their apparently only viable mode of transportation.

Insurers can take into account the overall trends in national risks trends for automobiles (eg new car safety) as well as local or hyper-local risks (eg hail damage in the southeast, property crime in a particular zip code). And as a large country with nearly as many cars as people, many insurers are willing to meet the demand. This satisfies criteria B and C.

So well-organized is the automobile insurance market that you could almost say that it’s vertically integrated: the largest nationwide insurers have contracts in place with every dealership network, auto collision chain, new and used parts dealers, as well as automatic data sharing with state DMVs, plus with firms like CarFax that buy information. Despite each state being slightly different, the insurers have overcome and achieved a level of near uniformity that allows an efficient market to exist.

Things are drastically different for the American healthcare system and for American health insurance companies. While most think of their healthcare provider as a national name like Anthem Blue Cross or Kaiser Permanente, the reality is that each state is an island. Even federal programs like Medicaid and Medicare are subject to state-level non-uniformities. Hospitals can be either privately operated (eg religion-affiliated, or for-profit) or run by a public entity (eg county or state), and can exist as a single entity or form part of a regional hospital network). Some entities operate both the insurance pool as well as providing the health care (eg HMOs like Kaiser Permanente) while others dispatch to a list of contracted providers, usually being doctor’s own private practices or specialist offices.

With so many disparate entities, and where healthcare is a heavily-regulated activity by each state, the cost of insurable risks – that is, for routine healthcare services – is already kinda difficult to compute. Hospitals and doctors go through intense negotiations with insurers to come to an agreement on reimbursement rates, but the reality is that neither has sufficient actuarial data to price based on what can be borne by the market. So they just pass their costs on, whatever those may be, and insurers either accept it into their calculations, or drop the provider.

Suffice it to say, there are fewer pressure to push the total cost of healthcare down, given this reality, and more likely prices will continue to climb. This fails criteria C.

financial flow in the US healthcare system Source

Briefly speaking, it’s fairly self explanatory why people would want health insurance, since the alternative is either death or serious health repercussions, paying out-of-pocket rates for service, or going to the ER and being burdened by medical debt that will somehow haunt even after death.. Criteria A is present.

As for Criteria B, that was actually resolved as part of the Affordable Care Act (ACA). During discussions with the drafters, insurers bargained for an obligation for everyone to have insurance (aka the individual mandate, bolstering criteria A), in exchange for an obligation to issue policies for anyone who applies, irrespective of pre-existing health conditions. Thus, Criteria B is present for all ACA-compliant policies in the USA, even though the individual mandate was legislatively repealed.

So to answer your question directly, the costs for healthcare in the USA continue to spiral so far out of control that it causes distortions in health insurance market, to everyone’s detriment. Specific issues such as open-enrollment periods, employer subsidies, and incomprehensible coverage levels all stem from – and are attempts to reduce – costs.

Enrollment periods prevent people from changing plans immediately after obtaining an expensive service, like a major surgery. Employer subsidies exist due to a federal tax quirk decades ago, which has now accidentally become an essential part of the health insurance and health care situation. And coverage levels try to provide tiered plans, so people can still afford coverage for “catastrophic” injuries while others can buy coverage for known, recurring medical needs.

But these are all bandaging the bleeding which is unchecked costs. It would take an act of Congress – literally – or of state legislatures to address the structural issues at play. The most prominent solution to nip costs is the bud is to achieve the same near-vertical integration as with automobile insurance. This means a single or very few entities which have contracts in place with every provider (doctors and hospitals), negotiated at once and uniformly, so as to achieve criteria C. The single-payer model – which Medicare already uses – is one such solution.

Going further would be the universal healthcare model, which discards the notion of health insurance entirely and creates an obligation for a government department to provide for the health of the citizens, funded by taxes. This means doctors and hospitals work at the behest of the department for then citizenry, or work privately outside the system entirely, with no guarantee of a steady stream of work.

These models could be approached by individual states or by the nation as a whole, but it’s unclear where the Overton window for that idea currently is.