Here are the facts. Chained-CPI does mean that Social Security beneficiaries will see their benefits cut. Imagine a person born in 1936 who retired in 2001, at age 65. For simplicity, let's assume they're eligible for the maximum benefit. Given that the cap was below $30,000 a year as recently as 1980, it's not inconceivable that a middle or upper-middle class person with steadily increasing earnings since 1958 would be in this situation.
Their initial benefit would have been $1,538 a month, or $18,456 a year. Under existing law, they would have gotten a series of cost-of-living adjustments (COLAs). By 2013, COLAs would have increased this person's annual benefit to $24,689.49. However, under chained CPI, it would be $23,820.19, a decrease of $869.30. That's a 3.5 percent cut in benefits. And, of course, a 3.5 percent cut in income matters a lot more when you're barely clearing $20,000 a year than it does when you're making a regular middle-class salary.
There would be other complications as well. Kenneth Stewart, an economist in the Division of Consumer Prices and Price Indexes at the Bureau of Labor Statistics (BLS), is one of the guys who computes the various CPIs every month. He notes that one benefit of CPI-W and other unchained CPIs is that they are final upon issuance. That is, the numbers are never revised. Two weeks after this month ends, BLS will release the April 2013 CPI, and that will always and forever be the April 2013 CPI.
Not so for chained CPI. "The chained CPI-U is subject to revision because we don't get the actual expenditure data until 1 or 2 years later," he notes. For example, in 2005 we had access to final chained CPI data for 2003, and only interim data for 2004. If we were to adopt chained CPI, we'd either have to use incomplete data, or else wait until we had final data to implement COLAs, which would further compound the cuts. The former, of course, would reduce the accuracy of the measure, a feature that proponents often tout.
For that reason, critics of chained CPI have sometimes promoted the CPI-E, an experimental index meant to measure price changes within products bought by the elderly. Because it's experimental and simply a result of reweighing the existing CPI measures to more heavily account for goods like housing and health care, the BLS doesn't publish the data on its website, but it's available upon request. Stewart, who helped develop CPI-E, explains that it's an unchained measure, and because of that its numbers don't need to be revised.
In the mid-2000s, the housing bubble and boom in health-care prices meant that CPI-E, which weights both more heavily, rose faster than conventional inflation measures. In 2008, for instance, adopting CPI-E as Social Security's inflation measure would have given our hypothetical retiree $327.88 more a year. However, since the housing bubble burst and health-care prices started slowing in growth, that effect has diminished. "Medical care inflation has been relatively subdued," Stewart says. "Shelter prices have also been very tame in, really, the last seven or eight years." As a result, in 2013 CPI-E would have resulted in only $56 in additional annual benefits for our test retiree.
The measure is not without its disadvantages. The biggest, noted by the Committee for a Responsible Federal Budget's Adam Rosenberg and Marc Goldwein here, is that a third of Social Security beneficiaries are not elderly. If the goal is to accurately represent inflation for beneficiaries, you'd also have to look at spending patterns among survivors and the disabled, which would be a considerable new project. Also, CPI-E uses a small sample size as it relies on the same sample that normal CPI does. To bring it from the experimental phase to become ready for prime time, that'd need to change. "We'd have to do a lot of research as to where those folks are shopping, what exactly they're buying, and what prices they're paying," Stewart says. "It's very difficult and costly to measure."
But ultimately, the question of which you prefer likely has more to do with whether you think Social Security benefits need to be pared back to ensure the program's long-run solvency, or whether you think the elderly need, if anything, a benefit bump. Those are policy questions, not technical ones, and all the debate in the world about chained CPIs and CPI-Es relative methodological merits won't resolve them.