Senator Edward Kennedy has just proposed simply expanding Medicare [1], in contrast to complicated plans like Wyden's and Schwarzenegger's. It's a good idea, but why it's a good idea isn't entirely clear on the face of it. What I'm going to do below is discuss the basics of how insurance works both from an actuarial and underwriting point of view. The principles aren't complicated, so I hope you'll join me for this explanation of one part of how insurance works.

One of my friends used to be an underwriter for a very high end life insurance company and she called the job "making book on people's lives." This is something many people don't quite get - when a life insurance company insurers you, for example, what they're betting is that you will live longer than you think - and they price you based on putting you into a risk classification for death. Very basic life underwriting is that men live less long than women; that smokers live less long than non-smokers and the old people live less long than younger people. So, all other things being even, a 70 year old male smoker will pay more than a 69 year old female smoker, for life insurance (the reverse would be true for health insurance, especially in places where companies are allowed to charge more based on gender.)

But a good underwriter will look at other things - how obese you are, what job you have, whether you engage in hazardous sports, how long your parents and grandparents lived, and will go through your medical file in detail. Then they slot you into a category on a chart. If done right, the average mortality rate for that group will be what is expected and the insurance company will make money - the insurance is calculated to bring in enough money to cover benefits, plus to pay profits, based on the expected deaths over time.

Now the larger the number of people you have enrolled the more reliable this is. Because this is about averages. A 70 year old female with a certain history might be expected to live 7 more years on average - but there's still a chance she could die tomorrow - or she could make it to a hundred. The larger the number of people in the group, the more like that the group's deaths will average out to what the actuaries expected when creating the chart and thus the more likely it is that the company will make what it predicted.

That is to say - the larger the group - the population - insured, the lower the volatility. If a small group is insured, then the costs go up, because the insurance company will put in an extra charge to cover the chance that people will die sooner than expected. Because if too many of them die early, the company actually loses money.

Actuarial charts are based on the experience of the population as a whole, then subdivided into different risk slots. The population as a whole (say, all 70 year olds) will almost exactly experience the expected mortality or morbidity (morbidity being the term for how often and how seriously people get sick). In fact for life insurance it'll experience slightly better performance, because the charts are done slightly conservatively, and people are living longer. For morbidity this is more variable - rates of certain illnesses keep going up, so you may well get worse morbidity than you expected. Nonetheless the larger the population insured - the lower the variation, the likely you are to get claims close to what you expected - and therefore the less the insurance costs.

The second issue that should be discussed is what's known in the industry as anti-selection. Put simply, people who are sick, or have good reason to think they're going to get sick are more likely to want health insurance People who have reason to think they might die sooner than later are more likely to want life insurance. This is a very real and measurable phenomenon.

If you insure the entire population - like Medicare - this goes away. There is no anti-selection - you get the entire population from age 65 and up, including all the healthy people and all the sick people and have experience almost exactly like your actuarial charts. Anti-selection isn't an issue.

The reason that private insurers like to insure through company insurance is related to this - a company is set up for a reason other than getting insurance. People are hired to do a job, not because they need insurance. If anything there is a bias against anti-selection - really sick people don't work. So the insurance company knows it is getting a group that doesn't have an anti-selection problem, that is in fact reasonably healthy. They can expect that while there may be some variation in the morbidity experience, it's unlikely to vary too much in the really bad direction. The bigger the employer, the better. Of course any employer will have less than the whole population of any age group, and will have more variability than if you were dealing with the whole population, but it'll have much less variability than a small group of self-selected individuals.

Individuals and companies are both underwritten. Underwriting individuals is about finding out if a person is likely to have bad claims experience (high morbidity for health insurance) and either charging them approximately what insurance costs, or excluding them if that's too much. If you have a serious pre-existing condition this is why you can't get insurance - it makes no sense. If you've got diabetes, the company knows it's going to cost a lot of money - to give you insurance they'd have to charge you that amount of money, plus their administrative costs, plus their profit margin. As an individual product it makes no sense. (Same thing with people with, say, terminal cancer. They don't get life insurance.)

All of this makes sense from a business point of view - people pay enough to cover the best odds of how much care they will need plus some extra for the insurance company (which is to say, the bookie.) It's a gambling operation and all honest actuaries and underwriters know it is. Just like betting on the horses, or sports, if the fix is in, there's no point in betting.

From a societal point of view the problem is that the model of insurance as gambling leaves the people who are obviously going to get sick out of the insurance system. People with preexisting illnesses can't get coverage. People like firefighters in California, for example, won't be covered by most health insurance companies because they're too prone to medical problems. You either charge them what it costs, or you have to spread the costs to other insureds and that makes their insurance more expensive - and your insurance uncompetitive with companies that refuse to accept them. If you're willing to just let those people suffer and die, then private insurance would make more sense than public - but if you believe that you can't just let those people go without the care they need, then it isn't and never can be. Which is one reason why the US actually spends more money per capita on health insurance than anyone else - because it spends money underwriting people, then pays for very bad health care for those the underwriting excludes, at very high costs, often when it's too late to the cheap intervention, and only the expensive ones can be tried (ie. when they stumble into emergency deathly ill and can no longer be turned away because they don't have insurance.)

Putting people into the right classification - making sure you don't get stuck with the people whose health is fixed, badly, and that you stick people into the right classification (it's a 3:1 bet on lucky lady to win the race) is what underwriting is about. It's not cheap, it takes a lot of people, and skilled people, to do it and a lot of data gathering (doctors offices often have a person or department whose only job is to make copies of medical records for insurance companies so they can review them.)

Underwriting is a big part of why private insurance has much higher administrative costs than Medicare. So is the fact that Medicare serves the entire population and thus has morbidity experience that is exactly, definitionally, average. Those two things are also why plans that keep private insurers in the mix, and still allow them to underwrite and pick and choose their applicants, are bad plans - because they will keep costs high.

And that's one reason why I think of all the plans that have been suggested, Senator Kennedy's plan to simply extend the number of people who can join Medicare is the best. Let kids join. Then decrease the age to 60, and so on. Medicare will always be cheaper than dividing up the population between private companies by the simple nature of insurance - the larger and more representative the population insured is to the general population, the more morbidity experience will be what you expect.

And it's the only way to cover people who have already lost the game - who are already sick, or who work at jobs that are likely to make them sick, or who have bad genetics or other risk factors.

Universal, single payor insurance is simply cheaper than alternatives if you aren't willing to let people who don't have insurance die without at least pretending to help them. And that's why plans which keep the market fragmented in order to keep the insurance companies in the primary insurance game will never be as good as those that don't. That's not to say there can be no place where private insurance wouldn't fit and be a good idea even in a single payor system - things such as top up insurance or platinum plated plans have their place and are even a good idea. For basic coverage, single payor is the way to go.

And that's why Kennedy's plan is the best of the three we've seen come out recently - Schwarzenegger's and Wyden's plans simply do not capture the efficiencies that single payor would and that's why if your first concern is getting good, cost-effective universal care to every American, you should put your heart behind Kennedy's plan and not behind plans which are pre-compromised from the beginning.