Jonathan Gruber of MIT has a new NBER working paper on health insurance coverage, based largely on a simulation model. Krugman cites the paper as evidence that making coverage mandatory makes a big difference for the number actually covered.
However, if you read the whole paper (unfortunately only available to subscribers), you find that Gruber sidesteps the crucial question of how a mandate will be enforced and just assumes a mandate will work. He writes:
The second option is ... by adding an individual mandate, a requirement that all individuals obtain insurance coverage, to the universal access option. This is similar to what is required for auto insurance in many states, and was a centerpiece of the recent Massachusetts reform plan. We have no experience to date with such a mandate, so it is hard to predict the success of enforcement. But, if penalties are strong (as they are in Massachusetts, where individuals are liable for half of insurance premiums even if uninsured), the mandate is likely to be close to universal. For simplicity here I assume that the mandate provides close to universal coverage, although in practice some individuals are likely to "slip through the cracks".This reminds of the old joke about the economist, the engineer, and the chemist trapped on a desert island, trying to figure out how to open a can of beans. After his colleagues offer elaborate technical solutions, the economist says "You're all thinking too hard. Let's just assume a can opener."
There's some similar logic at work in Gruber's analysis. Would a mandate ensure near universal coverage? Let's assume that it does! Question answered.
Since Gruber specifically mentions auto insurance, let's look at statistics on mandatory auto insurance as a very rough guide to how successful health insurance mandates would be. The Insurance Research Council produces estimates on auto insurance coverage rates; here is the press release from their most recent study, in 2006. Nationally 14 percent of drivers are uninsured, with 26 percent insured in Mississippi, and 25 percent uninsured in California and Alabama. (I couldn't find a listing of mandatory insurance laws by state, but insurance is definitely mandatory in the five states that have the highest rates of noncoverage.)
Thus, noncoverage rates for auto insurance are still very high, despite a wide variety of carrot and stick policies, e.g. requirements that proof of insurance be presented for registration, cross-checking between registration and insurers' databases, penalties for being caught driving without insurance, subsidized insurance for low-income drivers, etc.
This experience would suggest that a health care mandate, however enforced, would not be able to achieve "close to universal" coverage.
Of course you can achieve near universal coverage if the penalty for not having coverage is sufficiently draconian--say, imprisonment of those who don't sign up. But serious punishments will never be enacted in the first place because they would be perceived as too harsh. Clinton has tried to have it both ways with this question, by assuming 100% coverage with her mandate, but refusing to say how the mandate would be enforced.
All in all, this makes me think that my original conclusion--that the mandate question is largely a non-issue--is correct. See my earlier post on the topic for further discussion.