I thought commenter Eric made an interesting point that is worth quoting at length. He said:
Means testing social security seems a good idea on the surface. But isn't it ultimately rewarding bad behavior? And haven't we learned that rewarding bad behavior is not a good long term policy?
There are certain harsh realities to life. Is it possible that "people have to take care of themselves" is one of these realities?. Wishing it wasn't this way will not change it if it is.
I think there is one sense in which Eric is right. There is a way to design means-testing in a way that promote undesirable behavior. If Social Security benefits were designed to be paid based on assets and cash at the time of retirement, such a means-testing system would clearly and strongly encourage people in their forties and fifties to minimize their income and consume rather than save and invest. Higher earners would be able to have the purchasing power of the wealthy (since they would be turning more of their income into consumption) for most of their lives and a government-guaranteed retirement. A means-testing program in which benefits were means-tested to lifetime earnings would actually encourage greater saving and investment from high earners. A working life of upper middle-class lifestyle would only lead to a retirement of upper middle-class lifestyle if high earners save and invest.
Those who were lifetime low earners would get the "full" Social Security benefit but they would still have to work on the books to get the benefit. It is possible that low earners might work a little less than otherwise (though the incentive would be no greater than now for the vast majority) and save less in the expectation of the Social Security benefit, but let's put it in perspective: their reward for a working life of less-than-upper-middle-class purchasing power would be a retirement of less-than-upper-middle-class purchasing power. I'm not sweating the work incentives.
There are major work incentive impacts to not imposing some combination of means-testing and higher retirement ages to Social Security. The alternative of funding the existing system through taxes would mean some combination of higher payroll taxes on low and middle-income earners and applying the payroll tax to earnings above the current cap in order to give upper income retirees money they wouldn't need under a better designed system. Higher payroll taxes on low earners would tend to push low earners out of the formal economy and into off the books labor. Applying payroll taxes to earning above the Social Security cap would, combined with the forthcoming rise in income taxes for high earners and state taxes, push the effective marginal income tax for high earners well over 50% - which would really discourage work and investment. So you hurt low-income workers and hurt the overall economy in order to supplement the income of upper middle-class retirees. Makes sense to me.
Another way to go about it is to say no (or almost no) Social Security for anybody. The problem is that we seem to have a broad consensus against dealing with the exigencies of (ever longer) old age through rugged individualism (or rugged individualism plus family if you have any, plus charity.) That is why even more radical-sounding conservatives like Sharron Angle only argue for partially converting Social Security into a program of forced savings and investment rather than just leaving old people on their own.
The practical and political problems of means-testing are among the reasons I think that policy design and policy explanation is as important as the right general principles. It isn't enough to be in favor of "means-testing" or "privatizing." These kinds of reforms can be destructive if they are designed poorly. Supporters of such policies will need specific and concise answers to the obvious objections that will come from opponents, as well as explanations for the disasters that await us if reforms are not implemented