Here is a sentence you won't hear politicians or policy wonks saying in the next few weeks: "We should pay Social Security beneficiaries less in the future and push a lot of people into higher tax brackets." Here is a sentence you almost certainly will hear: "Let's adopt chained CPI."
Welcome to the dark art of obscurantist deficit reduction. Of course, the only ways to cut the deficit are by increasing tax revenue (either through higher rates, fewer deductions, or faster growth) or cutting spending. But both of those methods are unpopular. So, to get any support for their plans, politicians who insist on cutting the deficit have to find ways to cut spending or raise taxes that don't look like they're doing just that. Perhaps the most popular option along these lines is adopting "chained CPI."
Here's how it works: Numerous government programs, most notably Social Security benefits and the income thresholds for tax brackets, are indexed for inflation. But inflation can be measured in a number of ways. The tax code, for instance, uses CPI-U (Consumer Price Index - Urban), which measures prices for consumers in urban areas, to adjust the income cutoffs for different tax brackets. Social Security uses CPI-W, which is like CPI-U but only counts prices paid by urban wage-earners, not all consumers.
Various deficit-reduction frameworks, including Bowles-Simpson, Domenici-Rivlin and the Gang of Six plan, would convert all programs using CPI-U or CPI-W to a third measure -- called C-CPI-U, or chained CPI. Most inflation measures, including CPI-U and CPI-W, track the price of a certain basket of goods. That basket could include, say, a year's supply of propane. When propane costs go up, CPI-U and CPI-W include that as an increase in the cost of living.
But some people would just stop using propane if its price went up. They'd switch to electric heating, or a geothermal system, or a wood stove. So their actual heating costs wouldn't go up as much as CPI-U and CPI-W would suggest. Chained CPI attempts to take "substitution effects" like this into account. Thus, its number generally rises more slowly than other metrics.
That adds up to a big cut in Social Security benefits. Imagine, for example, a person born in 1935 who retired to full benefits at age 65 in 2000, with maximum earnings since age 22. According to the Social Security Administration, people in that position had an initial monthly benefit of $1,435, or $17,220 a year. Under the cost-of-living-adjustment formula and 2012 inflation, that benefit be up to $1,986 a month in 2013, or $23,832 a year. But under chained CPI, the sum would be around $1,880 a month, or $22,560 a year. That's a cut of over 5 percent, and more as you go further and further into the future:
The results by using chained CPI for taxes are also striking. The Tax Policy Center calculated the income tax increases that would be caused by a switch to chained CPI. They're not big — a little more than $100 a year for most families — but they're oddly regressive:
All told, chained CPI raises average taxes by about 0.19 percent of income. So, taken all together, it's basically a big (5 percent over 12 years; more, if you take a longer view) across-the-board cut in Social Security benefits paired with a 0.19 percent income surtax. You don't hear a lot of politicians calling for the drastic slashing of Social Security benefits and an across-the-board tax increase that disproportionately hits low earners. But that's what they're sneakily doing when they talk about chained CPI.
That's why watchdog groups like the Center for Budget and Policy Priorities argue that the only fair way to do chained CPI would be to pair it with an increase in Social Security benefits, and to exempt Supplemental Security Income, which provides support for impoverished elderly, disabled and blind people. Otherwise, it's just a typical "raise taxes, cut benefits" plan, and an arguably regressive one at that.
Update: The example retiree was initially presented as an average retiree. It is in fact someone with maximum earnings from 22-65. We regret the error.