That's a great story. Unfortunately, it has nothing to do with the debate over cutting Social Security. The problem with the Peterson story is there are very few wealthy elderly. We can zero out the benefits for Peterson and his wealthy friends and nothing in the Social Security accounts would change. Social Security's maximum benefits are just a bit over $30,000. Taking away $30,000 from a few thousand families makes no measurable difference in a program that spends $700 billion a year.
Getting a bit more realistic, let's suppose that we decide that $100,000 is the cutoff for affluent families. This is less than half of the $250,000 a year that President Obama sets as the cutoff for being subject to tax increases. A bit less than 10 percent of families over age 65, or 2.4 million, have an income of more than $100,000 a year. If we zeroed out their Social Security benefits, then we could save $70 billion a year, assuming they get an average of $25,000 a year from the program.
This 10 percent saving might seem a good start to reducing spending, but this requires a bit closer look. Suppose $100,000 is set as a cutoff, where you lose all benefits if your income crosses $100,000. This creates an enormous incentive to keep your income under $100,000. To be serious about this sort of affluence test it must be phased in gradually.
To the beneficiaries, taking away benefits as income rises is the same as taxing it. Seniors earning $100,000 a year will already be paying a 25 percent tax rate on their income. If we take away 25 cents in benefits for $1 in income, this would imply an effective marginal tax rate of 50 percent (25 percent plus 25 percent), not counting any state taxes that might apply.
We can argue as to whether this schedule is too high, but it implies that the government will save very little money from this affluence testing. Roughly half of our "affluent" elderly will fall in the range of $100,000 to $120,000. Assuming an average income for this group of $110,000, we will get an average of $2,500 from this group or a total of about $3 billion.
This schedule will get us a bit more from the richest 5 percent of the elderly, but we are unlikely to reduce the cost of the program by much over 1 percent through this sort of affluence testing. To have any substantial impact on the cost of Social Security we have to apply affluence testing to families that earn $60,000 to $70,000 a year, giving a whole new meaning to the word "affluence."
There is also a second arguably more important issue about affluence testing; these people paid for their benefits. They were taxed throughout their working lifetime for their benefits. The benefit structure is already highly progressive, so higher paid workers get a much lower return on their taxes than workers with lower wages.
These workers have a right to expect their benefits in the same way that someone who bought crop insurance from the government or who bought a government bond has the right to expect to be paid by the government, even if they happen to be very wealthy. It would be a serious breach of trust to take away benefits that workers had paid for, just as it would be a breach of trust to default on the bonds they held.
This is important not just as a moral issue, but also for political reasons. Social Security enjoys enormous bipartisan support because all workers pay into it and expect to benefit from it in retirement. Taking away the benefits that better-off workers earned would undoubtedly undermine their support for the program. This could set up a situation in which the program could be more easily attacked in the future.