I think you dismiss the "tax arbitrage" too easily. Between state and federal taxes, most people are looking at 30-40% marginal income tax rates. It takes a lot of inefficiency to overcome that kind of savings.

My personal view is that the tax breaks mean that companies pay for health care, which means that people don't pay for health care and then, of course, over-use it because they don't pay the cost. That effect causes the twin evils of over-priced health care (because the person paying for it is not the person receiving the benefit) as well as the farce of insurance companies deciding, without knowing anything about your personal situation, how much money to pay for any given medical condition.

One point you miss is that company benefits (at least for health care insurance) solve the "group problem". Insurance policies bought individually cost much more than group policies, so the $4000 policy my company buys with my $5000 foregone salary might well cost $8000 for me to buy from my own pocket. THis is partly due to higher sales and administration costs for individual policies but also because the company cannot gauge their risk as accurately.

I don't buy this "group benefits are cheaper" story. We can buy individual auto insurance--why not individual health insurance?

One theory is that in the self-employed market, young people don't buy health insurance. In the corporate sector, young people are forced to buy health insurance. So if you're middle aged, you want to go to the corporate sector, where you will be put into a "group" with young folks, thereby getting subsidized by them. I buy that argument, but it doesn't really lower the cost of health insurance--it just lowers the average amount paid for health insurance by forcing young people to obtain it.

What I was trying to say is that any reasonably sized company presents a smaller actuarial risk to the health insurer than an unknown individual who calls up wanting to take out a policy. This is more than just the cross-subsidy between age groups.

And I don't think auto insurance is a very good counter-example from the standpoint of social efficiency. Far too much of the premium dollars go to adminstrative and legal costs vs. actually reimbursing victims. Much as I dislike having the government involved in anything, I always thought there was a great deal of merit in Andrew Tobias' "pay as you drive" proposal.

"Any reasonably sized company presents a smaller actuarial risk.."

I don't see that. As an insurance company, I think I'd rather have 10,000 individuals than insure a corporation with 800 employees.

Risk pooling of individuals works fine in mortgages and credit cards. Credit card companies don't turn down individuals. Mortgage lenders don't turn down individuals. Somebody needs to explain to me what makes health insurance different.

As for per-mile auto insurance, if a company offered that, it would get killed. Every household with teenagers would flock to that company in order to get cheap rates, and it would have huge claims. So the only way to have that kind of insurance is with a government monopoly. And if you think that government monopolies are efficient...

In auto and home insurance, some people, depending on where and how they live, have more claims than others, but the claims process is largely memoryless. For each person, there is some probability p on a given day that a claim will be filed. Nevertheless, many people have a hard time getting auto coverage just because their personal value of p is a little high. On the other hand, sickness is a process with a long memory. Insurance companies would be highly motivated to cancel their very sick customers. Many people would become uninsurable as soon as time they get a long term illness. I suspect there is safety in collective company benefits in that the large company is far less likely to be dropped just because one employee gets a longterm illness.

Of course, this is just a hypothesis. Probably it could be tested by examining actual policy renewal data of those with individual insurance vs. those with company insurance.

You're arguing that because chronic illness is a useful predictor, future health risk is more predictable than, say, mortgage default risk. Not sure that I would agree, but let's stipulate that this is correct.

There are a number of solutions. You could have the insurable event be the diagnosis of a disease rather than its treatment. Or you could have health insurance be a long-term contract. If I am insured by ABC insurance in 2004, then ABC must cover any treatment for any illness I contract in 2004. So in 2007, even if I have a new insurance deal with XYZ, ABC pays for treatment for the chronic illness that I contracted in 2004.

In reality, I have individual family health insurance, we've had some really expensive chronic illnesses, and the company has only raised our premiums because we have grown older and because its overall claim experience has worsened as health care costs have risen.

Assuming your family's chronic illnesses imply that the expected cost of your policy is much higher than your premiums, why doesn't your health insurance company cancel you? Are they under a legal obligation to continue your policy? Maybe you should try an experiment and shop for new coverage?