Means-Testing Medicare

Robert Pear reports in today's New York Times about a little known feature of the 2003 Medicare Act: the premiums for Medicare Part B will now be an increasing function of a beneficiary's income. Part B is a voluntary supplement to Part A--the hospital insurance part that is covered by payroll taxes. Historically, the premiums for Part B have covered about 25 percent of program costs. The rest come out of the General Fund.

According to the article:

It is expected to affect one million to two million beneficiaries: individuals with incomes exceeding $80,000 and married couples with more than $160,000 of income. For individuals with incomes over $200,000, the premium, now $88.50 a month, is expected to quadruple by 2009.

[...]

When the transition is complete in January 2009, according to Medicare actuaries, the total premium for a person with income of $80,000 to $100,000 will be 1.4 times the standard premium. A person with income of $100,000 to 150,000 will pay twice the standard premium. A person with income of $150,000 to $200,000 will pay 2.6 times the standard premium, and a beneficiary with more than $200,000 of income will pay 3.2 times the standard amount.

If the basic premium rises 10 percent a year — a relatively conservative forecast — the most affluent beneficiaries will be paying premiums of more than $375 a month in 2009.

[...]

The Congressional Budget Office says 5 percent of beneficiaries will be affected.

[...]

The Congressional Budget Office estimates that the surcharge will raise $15 billion from 2007 to 2013.

The last excerpt suggests that we are not going to raise big money here. The reaction to the means-testing when it becomes known is likely to be as quoted here:

Theodore R. Marmor, a professor of political science at Yale, said the surcharge was more important for the politics of Medicare than for the financing of the program.

“The new income-related premium is fundamentally at odds with the premises of social insurance,” Mr. Marmor said. “Large numbers of upper-income people will eventually want to find alternatives to Part B of Medicare and will no longer be in the same pool with other people who are 65 and older or disabled. Congress will then have less reluctance to cut the program.”

It is very difficult to means-test a voluntary program. I think that problem is insurmountable here, as long as the means-testing is designed to affect only the top 5 percent or so. I am also not moved by the "fundamentally at odds" complaint. Enormous transfers from young to old under the guise of intragenerational redistribution from the healthy to the unhealthy are at odds with sensible fiscal policy.

When I got to CEA in July 2003, the Medicare bill was already in conference. I read the proposed rules for means-testing based on annual income of seniors and decided I had better get to work. I wrote a memo arguing that it was better to means-test based on lifetime income. I envisioned the tax base to be the average indexed monthly earnings of the individual (or couple), where earnings are defined as those on which the Medicare payroll tax was paid.

Since that memo stayed in DC, I outsource the commentary to Gene Steuerle (remembering him in my thoughts today), who made similar arguments in 1997:

An alternative way to apply the new fee in a progressive manner would be to base it not on annual income, but on the lifetime earnings of individuals, as reported to the Social Security Administration. Most of these lifetime earnings are currently applied to the calculation of something called average indexed monthly earnings, from which social security cash benefits are determined. The social security formula itself is progressive, and grants a higher rate of return to those with lower lifetime earnings.

There are three advantages to this approach. First, the Medicare beneficiary would easily perceive the charge as an additional fee for Medicare, not as something else. Second, the additional charge would be easily administered. Those with high lifetime earnings and higher social security checks simply would find that a higher fee was assessed against them for Medicare. They already pay the rest of their Medicare fee out of their monthly social security check, and the amount withheld is exact. No additional forms need to be filed, and no reconciliation is required at the end of a year. Next year, instead of all beneficiaries bearing an identical additional fee of say, $3, only those with lower lifetime earnings would see a $3 increase, but those with higher incomes would see a larger increase, say, up to $20.

A third advantage is that past lifetime earnings are not as easily manipulated as is annual income for those who are already retired. Like annual income, however, no one measure is perfect, and some adjustments would need to be made -- as they are in the case of social security -- for those who are measured as having low lifetime earnings because they worked in federal or other employment that was exempted from social security tax. While these adjustments would not be perfect, on average I would guess that they still would be more equitable than annual income as a base for assessing the fee.

The memo received a good hearing in the White House, and I even got to visit the conference committee and explain the ideas to the staffers. The ideas didn't prevail, though, largely because this method of means-testing does not include unearned income at any point during a person's lifetime in its defintion of the tax base. I thought that was a very minor point (particularly given the manipulability of unearned income in retirement) compared to the ease, transparency, and robustness of the alternative measure.