Means-testing for Medicare

I’ve several times advocated the means-testing as a way out of our current and forthcoming fiscal problems.  Greg Mankiw (see also knzn) offers the classic criticism:

…from the standpoint of incentives, means-testing is equivalent to a tax
increase.  As a result, economists worried about the adverse incentive
effects of taxes (like me) should be also worried about the adverse
incentive effects of means-testing.

The point, of course, is well-taken.  But something must be taxed, and Medicare benefits for the well-off are a logical candidate.  They represent the spending of relatively wealthy people, rather than savings.  In behavioral terms, I suspect the negative incentive effects of means-testing are relatively weak.  A person might say "If I get too rich, I’ll get less Medicare when I am old," and work less.  Dollar-for-dollar I expect this effect is weaker than "They’ll take out another few percent this year from my paycheck, maybe I’ll work less."

I’ll stick with means-testing as the least bad way of raising (implicit) marginal tax rates.  Means-testing also gets people away from the seductive but dangerous idea that government should take care of everyone, all the time.