Jonathan Gruber has just written a very useful and comprehensive paper on health insurance (I don’t yet see ungated versions). He estimates that without a universal mandate, but using subsidies, a typical plan for covering the uninsured would cost $4500-$5000 a year per person, and that is cost in the narrow budgetary sense. With a mandate the fiscal cost of the government (again, not social cost, which includes the cost of paternalistically forcing people to buy health insurance) is estimated at $2732 per person per year. Of course it is cheaper to tell people what to do, comparing to paying them to do it. That cost estimate is assuming that the mandate is effectively enforced, which I do not expect.
I would have preferred the primary estimates to be in terms of social cost. And I would have liked a discussion of how mandates and minimum benefit requirements distort the price of health insurance and limit competition. Read Shikha Dalmia. Nonetheless this remains is one of the best papers on health care economics to be had.
Gruber also poses an interesting philosophical question for the paternalists: would you rather be uninsured in today’s America or obese? And if you, like I, answer "uninsured," why not first direct paternalistic interventions toward obesity? And I’m not talking about subsidies to olive oil, I mean real mandates. After all, they will lower health care costs, no?