As my colleague Michael Cannon discusses below, in today’s WSJ Online, Carl Bialik examines the data on how many Americans do not have health insurance. Discussions like this one will be rehashed repeatedly during the coming health care debate, but they miss the crucial point: the U.S. should not expand government subsidy for health insurance whether the number of insured is 46 million or just 46.

The economics argument for subsidizing health insurance rests on the claim that private insurance markets do not provide fairly priced insurance. This is allegedly because insurers cannot distinguish the good health risks from the bad health risks and thus price insurance at a level only the bad risks are willing to pay.

This claim of “asymmetric information” is incredibly unpersuasive: absent regulation to the contrary, an insurance company can require any medical tests it wants and learn an insurance applicant’s health at least as well as the applicant. It can also condition coverage on relevant behavior, such as not smoking or maintaining a reasonable weight.

The problem is thus that insurance companies can determine all too well who is a good health risk and who is not, so they will price insurance accordingly if the law permits. This strikes many people as unfair, so they want to subsidize insurance for those born with unhealthy genes.

If insurance subsidies had few unintended consequences, this might be a reasonable form of social insurance. The problem is that subsidizing insurance exacerbates moral hazard, the tendency of people with insurance to consume too much health care. This is a crucial reason for rapidly increasing health expenditures.

Policy must therefore accept a trade-off: subsidizing health insurance will increase some people’s perceptions of fairness, but it will make the health care market less efficient.

A reasonable balancing of these two concerns suggests subsidizing insurance for the truly poor, but no more. In fact, the U.S. already does that via Medicaid. The uninsured are mainly people with too much income to qualify for Medicaid, or people eligible but fail to apply. Thus expansion of subsidized insurance to the currently uninsured, whatever their number, is likely to generate substantial inefficiency relative to any increase in “fairness” it creates.