Sunday, September 30, 2007

Fair and Efficient Health Care

The United States are about to overhaul—or perhaps apply an expensive patch to—their health care system. Given a blank sheet, what would be the fairest efficient universal health care system?

Lessons learned from Quebec's system
An effective way to limit costs without recourse to patient disbursement is to increase waiting times for both small and large interventions. Getting oral contraceptives prescribed requires waiting a couple of hours at the doctor's office. Getting knee surgery requires waiting six months. The value of one's time combined with hyperbolic discounting raises the effective cost of treatment far above 0$. Often, ills fix themselves faster than the health care system can, at some cost to the patient's comfort and productivity. Less frequenty, irrecoverable harm such as death occurs during the waiting period.

Anyone who has ever spoken to a human teller at a bank knows that waiting is not limited to public services. Without building significant spare capacity, queues arise naturally and delays follow a poisson distribution. Waiting is generally a bad thing: auctioning off the priority of access to medical services —not the access itself— could get around the poisson distribution problem by properly valuing people's time. But such an approach is quite unpopular, and not entirely fair. Charging a small flat cost could also avoid unnecessary cases, but this is still too controversial to be accepted. Lesson #1: limiting the supply of a medical intervention is not a fair or efficient way to limit costs.

The state runs all hospitals and sets prices for all private practice interventions. This monopoly position allows medical professionals' wages to be kept artificially low, which is quite unfair to doctors, but a good deal for the rest of us. Lesson #2: markets should set the wages of medical professionals, otherwise they are not getting fair compensation. State run facilities are inefficient in the administration of the services. Privatization of all medical facilities would increase the wages of the medical staff; this increase in wages may not be entirely offset by increased efficiency in the management of the facilities. Assuming that wages should be increased to free market levels, the quality of service offered to patients would increase dramatically at no cost. Lesson #3: privately run facilities deliver better service at a lower cost.

It is very difficult to sue your doctor in Canada. And in the unlikely event that you win your case, compensation is rather limited. This is a major difference between the Canadian and American system, which significantly reduces malpractice insurance fees paid by hospitals and practitioners. Lesson #4: keep medical malpractice insurance costs low.

Health related immigration is an issue. Uninsured foreigners come to Canada to be treated, sometimes as refugees, other times stealing the identity of Canadian citizens to do so. Lesson #5: immigrants should cover their health insurance costs.

Other practical issues
The insurance system always ends up being split into two: the baseline universal government sponsored insurance, supplemented by private insurance and self insurance. Human lives are too long to keep the division between what is covered by public and private insurance programs constant. Generally, public services are increased as productivity and technology improve. Such increases will force single risk private-insurance contracts to end prematurely, which will result in an overpayment (thus a refund) to the private insurance customer. Very exceptionally, costs may increase (eg. inflation caused by a receding population) or government budgets may decrease (eg. a depression). Good planning by the government is required to avoid decreasing services, as this is extremely unpopular. Decreasing government insurance coverage is the equivalent of defaulting on their insurance contract. For high risk persons, acquiring private insurance after a government default will not be possible. Lesson #6: the government should be conservative in the size of the baseline insurance package to avoiding defaulting.

Discrimination in insurance is a very delicate matter. Some things you can discriminate on, others you can't. Race and DNA tests are currently not acceptable ways of pricing private insurance. Should they be? The answers are not obvious. Of course if you offer identical and adequate health insurance to everyone, no discrimination whatsoever occurs (for better or worse). Discrimination is generally a bad thing when no incentives are at play. But some things you choose affect your insurability: should you choose smoking or lower insurance premiums? Should you risk giving birth to a child which will likely be unhealthy as shown by the results of an amniocentesis? To be fair with both the insuree and insurer, full information about risk should be disclosed to all parties prior to entering a binding insurance contract. If the insuree has more information than the insurer (as is the case now), high risk individuals will acquire more insurance coverage at the cost of others. Lesson #7: information used by the insurer and insuree to enter in an insurance contract must be identical.

Private insurance contracts may pay a fixed amount for a given illness, enough to cover the costs of the necessary medical intervention in most venues. What happens when some medical facilities charge more or less than the covered amount? Should the difference be payed or received by the patients? Would patients intentionally become sick to collect the difference as a source of revenue? If no medical assistance is required by the patient for an illness, should the sum be entirely given to him? Ideally, I would answer yes to all these compensation questions. Some costs have to be incurred in preventing, investigating and paying fraudulent claims. I hope these are not too high. Treating an illness quickly usually increases productivity, which is a collective good. Productivity loss is an externality imposed on others through reduced taxation and trade. Thus making health insurance claims redeemable in cash may not be optimal for society as a whole. Then again, it is the only way of putting downward pressure on the price of medical services. Lesson #8: achieving downwards pressure on price of a medical act is not possible without cash compensation paid directly to the patient. Cash compensation to the patient can be easily abused for some illnesses, and may lead to reduced overall productivity.

A theoretical solution
The really clean but not too practical answer to this insurance mess is to force parents to insure their unborn children for a lifetime of baseline health insurance in one lump sum, which pays fixed benefits for a each illness. You could decide to either to test parents' DNA or the fetus' DNA for accurate pricing. The baseline insurance would be subsidized by the state at a given amount, say 250K$. If the baseline insurance cost more than 250K$, parents would have to pitch in the difference. If the insurance cost less, they would get the balance returned to them with interest over some period of time (say 18 years). This system is fair and seems to work quite well. Immigrants would go through a similar flow, except they would be tested later in life and priced accordingly (and not subsidized!). All pricing is private and competitive, thus reflects actual costs of offering the insurance.

One problem could arise if an insurance company when bankrupt for some reason. You would end up with uninsured citizens, some of which are now individually uninsurable. The government would have to regulate insurance companies and underwrite them, avoiding loss of coverage.

A much larger problem is the uncertainty in determining how much the coverage will cost when it will actually be needed: in seventy years! At a minimum, the seventy year bond market would need to be created. It is not possible to reasonably price such long term commitments. Additionally, the baseline service set would need to be changed over time to include more medical conditions as they become cheaply treatable. Older citizens whose contracts were negotiated at birth would not include the same coverage as they younger compatriots. Modifying their original contracts to add coverage necessarily leads to discrimination based on medical history, again making some people uninsurable for services added to the package. Clearly the lump sum at birth approach is impractical as long as 70 year forecasts remain a shot in the dark.

So one important flaw in private insurance is the time-line: long coverage periods come at significant increase in cost and decreased flexibility. Even in the purely libertarian approach, where everyone is rich enough to buy insurance plans from birth without any financial help from the government, such contracts would provide poor long term health coverage. Lesson #9: private insurance contracts must be limited to short periods.

Regularly renegotiated private contracts would definitely prevent unhealthy people from rolling over their insurance contracts, thus defeat the entire purpose of universal insurance. Limiting the information available to insurance companies in pricing their policies is essential in avoiding this. Giving them only the data available at birth fixes this problem fairly well. Only mandatory baseline insurance can be priced this way, otherwise customers who know they are more at risk than their birth data implies would skew their insurance choice to their advantage. Some problems remain. If the data available at birth include the customer's DNA, improving DNA analysis may reprice individual risk even if the dataset at birth has not changed. Also, insurers with the best service would end up getting all the sick customers, rather than their fair share of sick customers. Lesson #10: there is no good way for private insurance to provide universal health coverage.

Elements of a practical solution
- Private enforcement and execution of medical acts.
- Low awards for malpractice lawsuits.
- Health taxes and coverage must be different per age group, and may vary over time (though this is not entirely fair).
- Compensation should be paid out per illness as much as possible, and not always matched to the cost of a medical act. The government should set such compensation to the price most venues offer for the service. This is a messy gray zone.
- Parents having children with high/low health cost should pay or be paid for the difference (not too popular, this one).

Immigrants should be required to have the minimum "immigrant baseline health insurance", similar or identical to the one provided by the state to native citizens of the same age. These immigrants should be exempt of the health tax burden of native citizens. The health case of individual immigrants is too complex and fragmented to be priced efficiently by the government. Only private insurance companies can do this efficiently. Also, immigrants could have the option of continuing to pay health taxes in their native country as a partial or complete substitute to private insurance, as long as it matches the minimum "immigrant baseline health insurance". For example, citizens of most western countries could be allowed to immigrate because their native health coverage is considered sufficient, though not necessarily a superset of the native insurance required by their host country. Upon emigration, native citizens should continue payment of the health tax to their native state in order to continue receiving the same coverage in an other country. Calculating accurate compensation between the individual and the state for early termination of the health insurance coverage is not easy (many factors are at play here).